GRE 23 Measurement & Methodology

total-item correlation

-a method for increasing reliability -measuring the correlation between each item and the average score of the scale -excluding items that have weak correlation with the total score

Greater than minimal risk research

-any study that poses more than a minimal risk to participants -goes through a thorough review by the IRB -requires consent form

Reliability and Validity

-for a measure to be valid, it must first be reliable (give us a consistent result) -a measure can be reliable, but not valid -it will give you the same score every time you use it, but it wont measure your theoretical construct -it measures something with no error, but it is not your theoretical construct -even if you establish that a certain measure is reliable (consistent across time, or internally consistent) you still need to establish its construct validity -reliability is a necessary but insufficient condition for validity

confidence interval

-measures the probability that a population parameter will fall between two set values. The ____ can take any number of probabilities, with the most common being 95% or 99% -i.e. estimating a population mean on the basis of our sample mean, and stating our level of confidence that the population mean falls within our estimate

inter-rater reliability

-reliability as consistency across judges; methods of assessing reliability by having two different rates judge a behavior -popular in observational research

Beneficence (cost-benefit analysis)

-researchers must make sure to maximize benefits to humanity/participants (beyond personal benefits) and minimize cost to participants in their research -when planning the research, one must consider the costs to participants and think of ways to minimize them (often times, the costs are necessary to ensure internal validity (i.e. deception). One must consider those and find a compromise that would satisfy both ethical and scientific standards

Cronbachs Alpa

-the most popular measure of reliability -you administer the measure to one group of people -you calculate the correlation between each 2 items -the average of these correlations is your reliability

minimal-risk research

-there is some risk of harm, but it is minimal and no greater than risks encountered in everyday life -no stress and no more than minor deception are involved -typically information sheet and not a consent form

Stages in conducting research

1- research question 2- theory 3- hypothesis 4-methods (data collection 5- statistical analysis 6- interpret and conclude

Threats to internal validity

1. History- any external event can be responsible for the results (i.e. a political or media event, changes in whether, etc.) -if little time is lapsed between measurements there is less concern about history 2. Maturation- any change participants go through with the passage of time can explain the results (i.e. participants can grow older, hungrier, tired, bored) 3. Testing- first measurement impacted the performance on the second -participants are "primed" with the concept, are more motivated and more knowledgable about the purpose of the study 4. Regression to the mean- extreme scores on first measurement tend to regress to the mean on the second measurement -a phenomenon that occurs because of errors in measurement - if the score turned out to be extreme, it could be true, but it could also be due to random error due to problems with the measure (by chance you get an extreme number). Because your extreme score could be due to chance, next time you measure it you are more likely to get a score that is less extreme 5. Subject related factors- when participants figure our what you are testing and answer accordingly 6-Experimenter related factors- when experimenters at in a way that drives participants to behave according to predictions - overcoming threats to internal validity- -The study must involve a control group that is equivalent to treatment group in all aspects but the treatment -people need to be randomly assigned to treatment and control groups 7. selection- the groups may differ because they had different people in them to begin with, regardless of treatment

Autonomy

1. Participants should be treated with respect, as autonomous people with an ability to make rational decision as to whether to partake in the study or not (the researcher does not know better) -therefore, if participants are indeed autonomous they should receive all relevant information regarding the research (goals, procedures and risks) at the outset of the study so theycould make a decision whether or not to participate (consent form) -children or people with weak capabilities of making decisions (psychiatric patients i.e) -a guardian consent is needed -coercion is a serious threat to autonomy; getting grades for participating, research in prisons

Costs of research

1. Responsibility- once a researcher conducts a study, s/he must take responsibility for everything that happens to the participants in the study --researcher must acknowledge their power in the situation 2- the researcher needs to make sure no harm is caused to the participants --participants should leave the study in the same way they entered it both mentally and physically --therefore, the researcher must analyze the physical and mental risks that might be caused to the participants prior to conducting the study What are the costs? 1. physical harm- i.e. administering a drug, depredating sleep, blood tests, etc. 2. Mental distress- psychological anxiety and distress that participants might encounter in a study -negative feedback -expecting an electric shock -insults -remembering a traumatic event 3. loss of privacy and confidentiality- problem in longitudinal research when records of the same person should be matched How to minimize costs? 1- choose techniques that would cause the least physical and psychological discomfort 2. Maintain the anonymity of responses (also in cases where you need to identify the subjects) 3. obtain the consent of the participant 4. debrief the participants at the end of the study to make sure they leave the study in the same way they entered it -during the study, researchers should be sensitive and empathic -participants should be allowed to stop participating at any time -if the study is distressing, support should be available -the researcher should allow the participants to discuss the study and anything that bothered him/her at the end -rely on existing published methods which have already been approved by ethical committees and should be reliable and valid

establishing causality between two variables

1. The two variables must covary (As one changes the other changes) -note: this is also true for establishing a correlation 2. The cause must precede the effect -IV must appear prior to the DV 3. Alternative explanations must be ruled out (3rd variable problem must be overcome)

Benefits of research

1. Theoretical contribution- findings would contribute to the understanding of a phenomenon and would advance scientific knowledge in the area 2. Practical contribution- findings would contribute to advancing the well-being of individuals and would serve as a basis for developing more efficient methods for dealing with problems -in grant applications there is a special section dedicated to describing the theoretical and practical importance of the research - the idea is that if the study has no benefits it is better not to run it because then it would only bear costs To maximize benefits: 1- the study needs to deal with an important question in the area of research --one needs to provide a literature review that is updated and thorough and convince that the research question at hand is relevant and important -the researcher should avoid dealing with esoteric issues that have no theoretical anchor -the researchers must seriously consider the practical implications of the research and think how it contributes to society

Milgram study: ethical aspects

1. participants were deceived as to the exact nature of the study for which they volunteered, and making them believe they were administering real electric shocks to a real participant 2. Participants were not informed about the potential stress and emotional conflict they would experience 3. It is possible that being involved in the experiment may have had a long term effect on the participants. Their view of themselves as harmless human beings were shattered 4. The prods used suggested that withdrawal from the experiment was not possible

controlling the study setting

1.Making sure all participants go through the exact same experience, while only varying the manipulation 2.Variables that are held constant cannot explain the results 3.Confounding variables 4. The IV should be the only variable that differentiates between the groups in that study

Positively skewed distribution

A distribution is ___ if the scores fall toward the lower side of the scale and there are very few higher scores. ___ data is also referred to as skewed to the right because that is the direction of the long tail end of the chart. -scores are bunched up at the left and tail off (or skew) to the right -central tendency sequential occurrence in a ___ distribution is mode, median, mean (in that order)

Example

A researcher is exploring the relationship between eating and sleeping. specifically, the researcher hypthesis that eating within one hour of going to bed causes you to sleep unsoundly. The experimental group will eat just before bed, but the control group will eat 4 hours before bed. All subjects will eat and sleep in the lab. The IV is eating before bed. the DV is the quality f sleep. One potential confounding variable is the possibility that simply sleeping in the lab could affect the sleep of some subjects

Normal distribution (Bell curve)

Central tendency, variability and z scores are intricately related to the concept of ___. it is possible to speak in terms of probabilities, in a ___, -approx. 34% of the scores occur between the mean and a z score of +-1 (one SD above or below the mean). -Appx. 14% occur between z scores of +-1 and +-2 (between one and two SDs above or below the mean) -appx. 2% of score in the distribution occur beyond a z score of +-2 So, if we refer to distribution 1 (distribution 1 has a mean of 40 and an SD of 2.5), momentarily and ask what the probability is of a score between 42.5 and 45, we can immediately state that probability as 14%. Because a ___ is symmetrical, exact percentages hold for scores occurring below the mean. i.e., the probability of a score between 35 and 37.5 in distribution 1 would also be 14%. -in an ideal world, scores such as those for the GRE are intended to look like a bell curve. The larger your sample, the greater your chance of having a normal distribution of values. 1- it is unimodel- it only has one hump. The majority of scores fall in the middle ranges. There are fewer scores at the extremes. The mean, median and mode are all equal in a normal distribution. 2-Z scores- 3- T scores 4- standard normal distributions

multitrait-mulimethod technique

Donald Campbell and Donald Fiske created the ____ to determine the validity of tests

Step 4: Compute statistics

Having selected the experimental design in Step 2, and the significant level in step 3, you should find this fourth step self-explanatory. Depending on the design that you have selected, you may have a t-value, z-value, f-value and so on. In each instance, for the n or df in your experiment (based on the number of subjects you have in each group) you will refer to the appropriate table (z, t, F) to determine whether the value you have obtained is larger than the value required for significance at the level selected

Step 5: Make a decision

If the number obtained in your computation is larger than the number found in the table for your significance level, (at your appropriate n or df), you can reject the null (no difference) hypothesis. As mentioned earlier, every experiment hopes that the difference found will be large enough for such rejection of the null. On the other hand, if the number you find in computation is smaller than the table value for your established significance level, you have failed to reject the null (meaning in this case that carrot eating did not have a significant effect on reading speed). Obviously, significant results are the prime candidates for publication in experimental-professional journals.; and the findings most impressive to journal editors are those involving experiments in which a large number of subjects have been used. Relating all this to your specific situation, suppose that you selected the .05 significance level and obtained a z score of 2.05 in your statistical computation. As you move to the appropriate column in the z-table (normal distribution table) you find that the score required for significance at the .05 level is 1.96. Since your score is larger, you can reject the null hypothesis and conclude that carrot eating has a significant effect upon reading speed. The table reference will change as a function of he statistical procedure that you have selected, but the basic reference and decision-making procedure will remain the same.

Point-Biseral Correlation

If, in the above described correlation setting, one of the scores you obtained was dichotomous, you would need a ___ to conduct the correlation. dichotomous suggests either-or in contrast to a score continuum. If you compare IQ scores with whether a person obtains an above B or below B GPA in college, the latter situation is dichotomous In tabling the dichotomous situation for purposes of the correlation, you might want to represent the above B performances by the number 1 and the below-B by 0. The IQ scores can occur in a large, continous range, and therefore your comparison would contain one continuous and one dichotomous measure for each person. Otherwise, the basic format would resemble the one that you would establish for the pearson product-moment correlation.

Chi Square

In discussing the nonparametric area of statistics, we mentioned that __ was one of the most prominent methods. ___ seeks to determine whether two variables are independent in a population from which a sample has been obtained __ deals with variables that are discrete categories rather than continous measurements, i.e., this statistic might be used to determine whether the variables of political party registration and sex are related. In the simplest ___ settings, you would be working with two categories for each of the dimensions (in this instance, female-male and democratic-republican). The wuestion is whether sex and political party affiliation are related in the population from which the sample was drawn. Because there are two discrete categories on each of the two variables, the resulting table represents a square, four paned window. In the procedure itself, you will obtain a value known as phi coefficient (similar to correlation coefficient) which can then be used to obtain a final __ value. includes terms such as: 1-expected frequency 2- obtained frequency 3-degrees of freedom -are used when the n-cases in a sample are classified in categories or cells. The results of the __ test tells us whether groups are significantly different in size. __ look at patterns or distributions (not differences between means). i.e. imagine that 100 members of an introduction to psychology class are categorized based on race (caucasian, african-american, asian-american, hispanic and native american). Insignificant results of a __ test of this data would tell us that no one race tended to be enrolled in introductio__ tests analyze categorical or discrete data (data that has been counted rather than measured and so is usually limited to positive and whole values). and can be used on small samples. ___ tests can also assess the "goodness of fit" (=the extent to which the observed data match the values expected by theory) of distributions or whether the pattern is what would be expected

Independent variables

In our example above, the ___ would be carrots- the stimulus element that is placed in the experimental situation to see whether it makes a difference in reading speed. -experimenter puts this variable into the design with the experimental group (as distinct from the control group) - the researcher is interested on the effect of the __ on the dependent variable. The researcher manipulates the __ often by applying it in the experimental or treatment condition and withholding it from the control condition

Step 3: set a significance level

In psychological research, the significance level is generally either .05 or .01. The .05 indicates that you are willing to consider a signifcant difference that can occur by chance only 5 times in each hundred cases. The 0.1 level is more stringent, accepting as significant difference that could occur by change only one time in each hundred cases. Notice that the significance level is set before statistics are computed. The sequence is essential. Otherwise, an experimenter might decide after the fact which significance level to choose- the decision then being based on the size of the difference actually found. Throught your persual of psychological literature, you will find expressions like "significant at the 0.1 level" or "significance at the 0.5 level" -Typically set at 0.5 (in effect, risking that 5 times in 100 ill make a type I or type II error) -more stringent=0.1 (risking only one time in 100)

Step 2: Collect the data sample B. Decide on a statistical procedure and collect data in a form compatible with that procedure

One of the most deplorable and traumatic scenes that any statistician can relate involves the sight of someone on the doorstep who has collected a batch of data and now wants to know what he can do with it statistically. Such decisions must be made BEFORE the data is collected.

Spearman Rank-Order correlation

Suppose that two judges were ranking the entries in a dog show. Judge 1- A-1 B-2 C-3 D-4 E-5 Judge 2-A-5 B-4 C-3 D-2 E-1 By comparatively scanning the above rankings, one can see that the dog ranked highest by Judge 1 was ranked lowest by Judge 2, that the dog ranked next to highest by Judge 1 was ranked next to lowest by Judge 2, and so forth. There is definitely a systematic relationship in a negative direction. ___ s rho formula yields a correlation coefficient of -1.0. f the rankings in Judge 2s column had been reversed, (1 for dog A, 2 for dog B, and so on) there would have been a perfect positive relationship between the judges rankings, and the resulting ___ rho coefficient would have been +1.0. -nonparametric version of the pearson product-moment correlation. ___ (p, also signified by rs,) measures the strength and direction of association between two ranked variables. Assumptions of the test: you will need two variables that are either ordinal, interval or ratio. Although you would normally hope to use a pearson-product moment correlation on interval and ratio data, the ___ can be used when the assumptions of the Pearson correlation are markedly violated. however, ___ determines the strength and direction of a monotonic relationship between your two variables rather than the strength and direction of a linear relationship between your two variables, which is what Pearson correlation determines

Percentile Rank

Take a piece of paper and put it over one of those bell-shaped curves we just talked about. Now slowly move the paper to the right, little by little, exposing the left side of the curve. What your doing now relates to ___. Percentiles begin at the far left of the distribution- a zero assumed at the far left. As your paper moves father and farther to the right, the percentile rank steadily increases. When you get to the midpoint (mean median and mode of this normal distribution) you have reached the 50th percentile. If your score came at the 50th percentile it would mean 50 percent of all those who took the test scored equal to or below you. As you continue to move your paper father to the right, youll reach 60th, 70th, 80th and 90th percentiles. i.e. the 95th percentile- this means 95 percent of all those who took the test scored equal to you or below you. -zero at the left side of the normal distribution, gradually increasing as one moves to the right -used most commonly on standardized tests. Along with your reported score of 750, you would receive a percentile rank of 97%. This shows your position in the whole group by saying that you scored higher than 97% of your group

Non-experimental methods

The variables are observed (and measured) as they occur naturally (the researcher has no control over the occurrence. i.e. - self-reports (how many times do you exercise # of visits to gym) -existing records (GPA and admission to grad school A relationship between the variables is established when the two vary together (covary--> correlation. i.e. more exercise report related to lower anxiety levels

Complex designs

There are two ways of making a design more complex 1- add more levels to your existing IV 2- add another IV -designs with more than one IV are called factorial designs In a factorial design, all levels of one IV are combined with all levels of the other IV -a 2x2 design will have 4 groups -a 2x3 design will have 6 groups -a 2x2x3 design will have 12 groups

Pearson Product-Moment Correlation

This correlation procedure applies in situations where the researcher wants to determine whether there is a relationship between two groups of paired numbers. Pairing generally means that two scores exist for the same person, thus in a typical situation utilizing this procedure, you would expect to have two sets of scores for each of several individuals, and would not want to determine whether the scores were in any way related. To illustrate, imagine that you have just obtained IQ scores and foreign language proficiency scores for a group of college sophomores. The question now arises of whether there is any relationship between intelligence and foreign language proficiency. To answer the question, you conduct a ___ on two sets of scores. If the resulting correlation is in the +0.6 range or above, there would appear to be a high degree of relationship between these two factors. You can begin to see how many factors and aspects of personal and social life can be examined with this method. i.e. correlations have been made between high school and college performance levels, and obviously correlations have been made between performance on the GRE and success in the graduate program -correlation does not mean causation If you think for a moment, you will realize that this correlation involves the same basic setting described for the use of t-test for related measures (Dependent t tests). The difference is that in the case of t-test you are comparing the same measure (taken at two different times or in matched groups) and are looking for a significant difference instead of or a systematic relationship

standard normal distributions

To combat problems of comparing scores and distributions of scores with different SDs, normal distributions can be standardized. There are known as ___. The ___ is the same as a normal distribution, but it has been standardized so that the mean for every such distribution is 0 and the SD is 1. ___ and z-scores allow you to compare one persons scores on two different distributions. i.e. if a persons intelligence matches his or her achievement on the GRE psychology test, you would expect someone to receive the same z-score on an IQ test and on the GRE psychology test (i.e., a z-score of +2), even though both tests use two different scoring systems. This would mean that the persons score should be 2 SDs above the mean in both cases. If we saw that the persons IQ score was 130, or two SDs above the mean, giving her a z-score of 2, and his or her GRE psychology test score was 570, or right at the mean, giving a z-score of 0, we would hypothesize that for some reason this individual was not performing to his or her potential.

Step 2: Collect the Data sample A.Set up the experimental and control conditions

To compare the responses of your experimental and control groups, you should administer carrots to the experimental group and no carrots to the control group (carrots-IV, reading speed DV) Because the control group will not be receiving the independent variable, this group will enable you to determine later to what extent an observed response change was a function of the carrot eating (IV) within the experimental group. Obviosuly, it is important to keep all other potential variables between the two groups the same. i.e. you may want to use only girls or boys to remove the possibility of performance differences resulting from sex differences. The reading material that you select must be equivalent for the two groups, and you must be certain that there hs been no previous familiarity with this material. It will be essential to measure the reading speed of all subjects BEFORE you institute the independent variable in order to get an accurate measure of any changes in reading speed after the experimental group has eaten their carrots, and it will be important to make sure that each person consumes the same amount of carrots. In addition, situational variables must be controlled-lighting must be equivalent for all subjects, and so on.

Parameters

Values obtained from populations - a descriptive measure of the population

Bimodal distribution

a continuous probability distribution with two different modes. These appear as distinct peaks in the probability distribution - the graph has two distinct humps or peaks with a valley seperating them. The prefix bi means two, so a graph with two peaks is called ___. Values in this distribution are either 0 or 1.

ANOVA (analysis of variance)

a highly utilized test because of its flexibility. It is similar to the t-test in that it analyzes the differences among means of continuous variables but it is more flexible than the t-test because it can analyze the difference among more than 2 groups (even if the groups have different sample sizes)

standard deviation

a measure of difference from the mean of a distribution (if high, its a widely scattered distribution, if low, its bunched closely around the mean) -square root of the variance. __ tells you the average extent to which scores were different from the mean. If the average __ is large, then scores were highly dispersed. If the ___ is small, then scores were very close together. Different standard distributions make it difficult to compare scores on two different tests.

internal consistency reliability

a method of reliability in which we judge how well the items on a test that are proposed to measure the same construct produce similar results. i.e., a question about the internal consistency of PDS (test for PTSD) might real "how well do all of the items on PDS, which are proposed to measure PTSD, produce consistent results?" if all items on a test measure the same construct or idea, the test has ___ reliability. i.e. suppose you wanted to give your clients a 3-item test that is meant to measure their level of satisfaction in therapy sessions. 1-"you almost always feel satisfied after your therapy sessions" 2-'you almost always enjoy therapy" 3-"you almost never feel satisfied with therapy If a client agrees with the first two items and disagrees with the third, your test has good ___ -tested with cronbachs alpha

Meta-analysis

a method of study that mathematically combines and summarizes the overall effects or research findings for a particular topic. Best known for consolidating various studies of the effectiveness of psychotherapy, meta-analysis can calculate one overall effect size or conclusion drawn from a collection of different studies. This method is needed when conflicting results are found and when different studies use different methods

lexical decision task

a procedure which involves measuring how quickly people classify stimuli as words or nonwords using priming For example, when you see the word "GIRL", you respond "yes, this is a real English word", but when you see the letters "XLFFE" you respond "No, this is not a real English word".

monotonic relationship

a relationship that does one of the following: 1- as the value of one variable increases, so does the other variable 2- as the value of one variable increases, the other variable decreases.

T-scores

a transformation of a z-score, in which the mean is 50 and the SD is 10. Thus the formula for calculating is T=10Z+50 The general rule of thumb for when to use a ___ is when your sample size meets the following two requirements: 1-The sample size is below 30 2-The population standard deviation is unknown (estimated from your sample data) In other words, you must know the standard deviation of the population and your sample size must be above 30 in order for you to be able to use the z-score. Otherwise, use the ___

Confounding variable

a variable that is not the IV and can serve as an alternative explanation for the results because it varies with the IV -the researcher attempts to minimize or eliminate confounding variables. These are variables in the environment that might also affect the dependent variable and would blur the effect of the IV on the DV

stratified sampling

aims to match demographic characteristics of the sample to the demographics of the population (i.e. sample that is 50% female, like the population)

nominal variables

allow for only qualitative classification. That is, they can be measured only in terms of whether the individual items belong to some distinctively different categories, but we cannot quantify or even rank order these categories. i.e. we can say that two individuals are different in terms of variable A(i.e. they are of different race) but we cannot say which one has more of the quality represented by the variable. Typical examples of ___ are gender, race, color, city, etc. -in effect, labels. the fact that your house number is 1054 and someone elses is 1020 does not mean that your house is beigger than the other persons. The numbers simply serve a labeling, categorizing function -comes from the latin word for name, so these variables are simply given descriptive names. There is no order or relationship among the variables other than to seperate them into groups. i.e. male, female, republican, democrat

Interval variables

allow us not only to rank order the items that are measured, but also to quantify and compare the sizes of differences between them. i.e. temperature, as measured in degrees F or C constitutes an interval scale. We can say that a temperature of 40 degrees is higher than that of 30 degrees, and that an increase from 20 to 40 is twice as much as an increase from 30 to 40 -to express consistend interval values at different points in a distribution (i.e. the difference between 15 and 20 degrees F is equivalent to the difference between 85 and 90 degrees F. -capable of showing order and spacing because equal spaces lie between the values. These variables, however do not include a real zero. i.e. temperature is ordered, and the values are equally spaced. So 75 degrees is 35 degrees warmer than 50 degrees. Temperature has an arbitrary zeroo, however there is no point that signifies the absence of temperature

Ordinal variables

allow us to rank order the items we measure in terms of which has less and which has more of the quality represented by the variable, but still they do no allow us to say 'how much more' a typical example of an ___ is the SES of families. i.e., we know that upper-middle class is higher than middle, but we cannot say, i.e. 18% higher. Also, this very distinction between nominal, ordinal and interval scales itself represents a good example of _. i.e, we can say that nominal measurement provides less information than ordinal measurement, but we cannot say how much less or how this difference compares to the difference between ordinal and interval scales - use of numbers involves rank ordering. Judges at the county fair use numbers in this manner. a given ___ can indicate more of a quality than another number, but it does not indivate that the distance between first and second, i.e, is equal to the distance between second and third. ___ is rank ordering of some characteristic that does not go beyond that ordering to any suggestion of equal intervals between. -implies order. Here, variables need to be arranged by order and thats it. Nothing else can be known because the variables are not necessarily equally spaced. i.e. marathon finishers- a different runner comes in first, second and third but we do not know how far apart their finishing times were, could be seconds or minutes

Linear regression

allows you to use correlation coefficients in order to predict one variable y from another variable x. Correlations measure the linear relationship between two variables, but they do not describe the relationship. i.e. we might know that the correlation between extraversion and number of friends is .73; but this does not let us predict from an extraversion score how many friends someone would probably have. ___ allows us to define a line on a graph that describes the relationship between x and y. In general, the same data you used to calculate a correlation is now plotted onto a graph. Lets use extraversion score as our x variable and the number of friends as our y variable. Imagine a graph with extraversion on the x-axis and number of friends on the y axis. The dots on the graph are the data from the individual subjects. ___ is when the least squares line, or regression line, is fit to the data and the line is as small as possible (this is determined by finding the difference between each data point and the line, squaring those differences to get rid of negative values, and then summing them). Fortunately, computer programs do this for you, and it will not be required on the GRE test. Just know that regressions use correlational data to make predictions based on a line fit with the least-squares method. -line drawn among a group of correlation-distribution points to represent a trend (i.e. how much y varies when x varies by one unit)

t-tests

among the prominent testing procedures are t-tests: 1- for the difference between sample and population means (1 sample) 2- t tests for two independent means (independent) 3- t tests of related measures (dependent/paired) -compare the means of two different groups to see if the groups are truly differnt. this would mean that the difference between the means is large enough to be considered statistically significant rather than due to chance variation. ___ analyze differences between means on continuous data (anything that is measured such as height or depression score on a depression scale as opposed to things that are counted such as group size, number of hospital visits, number of symptoms, which is discrete data) and are particularly useful with samples that have a small n (meaning few subjects). __ annot test for differences between more than 2 groups

mean

an average of the scores in the sample. -the same as the average. the __ of a set equals all the values added together divided by the number of values. __ are highly effected by extreme scores.

Latin square design

an experimental design that can be used to control the random variation of two factors. the design is arranged with an equal number of rows and columns, so that all combinations of possible values for the two variables can be tested multiple times. This design is used to reduce the effect of random or nuisance (i.e. loud noises, odors, traffic, etc.) factors

item analysis

analyzing how a large group responded to each item on the measure. This process weeds out dd or problematic questions so they can replaced with better questions (ones with discriminatory value)

Spearman r correlation coefficient

another correlation used only when the data is in the form of ranks. It is the procedure for determining the line that describes a linear relationship.

domain-reference tests

attempt to measure less-defined properties (such as intelligence) and need to be checked for reliability and validiy

Two way ANOVA

can test the effects of two independent variables or treatment conditions at once

between subjects design

compares 2 groups of people at the same time point -in a ___ different participants are assigned to each condition before the study -____ factor- different participants are assigned to each level of the factor (IV)

reliability of personality assessment

compute the correlation between self-report and informant report (someone close to the subject) -a common way to assess the reliability of personality tests

Belmont Report

defines the ethical principles that are to guide behavioral and biomedical research three basic principles 1. beneficience (cost-benefit analysis) 2. Autonomy (respect for persons) 3. Justice Institutions have an IRB that is responsible for reviewing research for violations of ethical standards

Platykurtic distribution

does not imply that the __ is flat-topped as sometimes reported. Rather it means the distribution produces fewer and less extreme outliers than does the normal distribution,

Order effects

due to the order of treatments, participants respond differently

Justice

equality and fairness towards participants regardless of age, ethnicity, gender or other criteria i.e. avoiding benefits from disadvantaged groups or even ust opening the study only for certain groups

two tailed tests

evenly divides the significance level on both sides of the distribution (i.e. if .05 level, .025 of it will be to the left and .025 will be on the right

1 sample t test

for the difference between sample and population means- i.e. a situation in which the researcher knows the national average for the dimension being measured (i.e. average weight of 12 year olds) and now must determine whether the weight of 12 year olds in the sample is significantly different from the population mean (the national average)

example of within/between subjects

if researchers wanted to see if a new drug decreased depressive symptoms, the researcher could use either design. - With a within subjects design, the researchers would measure each persons depression at baseline, administer the drug, and measure depression again. If the participants were less depressed the second time, the researchers would conclude that the drug worked.. -with a between subjects design, researchers would give one group of depressed people the drug and one group of depressed people the placebo and then measure depression. If the group that received the drug was less depressed than the group that received the placebo, then the researchers would conclude that the drug was affective

Z scores

imagine that distribution 1 has a mean of 40 and an SD of 2.5, while distribution 2 has a mean of 40 and a SD of 2.0. Now, if someone were to ask how a score of 45 in distribution 1 would compare with a score of 46 in distribution 2, a quick score comparison would be difficult to make. The computation of __ is a way of translating these different SDs and different means into a common language that facilitates comparison. Computed as score minus the mean divided by the standard deviation x-mean/SD the __ makes scores comparisons quite simple. In the above example, for distribution 1 a score of 45-40/2.5=+2. In distribution 2, a score of 46-40/2= +3. Therefore, through the use of the ___ translation, it becomes easy to see that the score of 46 in distribution 2 is significantly better than a score of 45 in distribution 1. Note that the __ in these instances was preceded by a plus sign. If the score had been below the mean, it would have been preceded by a minus sign -refer to how many SDs a score is from the mean, for practical purposes, z scores of normal distributions range from -3 to +3, because this covers the vast majority of scores on a normal curve

central tendency

in any distribution, it becomes necessary to measure ___. Most statistical procedures rely on the mean, but there is also the median and the mode. Use of the mean can be problematic in distributions where there are a few extremely divergent scores. Because it is an average, it tends to be prominently influenced by these extreme scores. In such instance, the median as the midpoint score makes a more appropriate measure of ____ What happens to the mean of a distribution when a fixed number is added to each score in that distribution? The answer is that the mean value is increased by this fixed number. In similar fashion, if each score was multiplied by a fixed number, the resulting mean would be the original mean multiplied by this fixed number. Division would have a similar effect. In summary, the same effect that has occurred with the individual scores in the distribution also occurs with the mean

Dependent variable

in our example above, the ___ would be reading speed- the response obtained in relation to the stimulus element introduced -subject response (and a comparison of differences in response between subjects in the experimental and control groups) -the researcher does not control the __, but rather examines how the IV affects the ___

Measures of central tendency

indicates where on a number line the data set falls in general. Three types of central tendency can be calculated on a set of numbers. mean, median and mode

exempt research

invloves no risk-anonymous questionnaires, naturalistic observations when there is no threat to autonomy -no consent is needed -the IRB determines if the study is exempt

Pearson r correlation coefficient

is a way of numerically calculating and expressing correlation. You dont need to know how to find a ___, but you do need to know what the r values mean. The ___ values range from -1 to +1. A value of -1 indicates a perfect negative correlation. A value of +1 indicates a perfect positive correlation. A value of 0 indicates no relationship. The strength of the relationship is indicated by how far away the value is from zero and how close it is to -1 or +1

Split-half reliability

is measured by comparing an individuals performance on two halves of the same test (odd vs even questions for example). This reveals the internal consistency of a test. Another way to increase internal consistency of a measure is to perform item analysis. -you administer the measure to one group of people -create two groups of items (randomly) -obtain 2 scores for each person; score for the 1st half (average of all items) and score on the 2nd half (average of all items) -high correlation between the scores on two halves indicates high reliability

Test-retest reliability

is measured by the same individual taking the same test more than once. On a test with high test-retest reliability, that person would get appx. the same score each time -we can test the same group of people twice, and examine the correlation among the measurement. -if people score similarly in both times (r larger than 0.7) it means the scale is reliable -common method for assessing reliability of 1-item scales -if people score very differently in both times than you have poor ___ of the measure disadvantages: 1-practice effect-participants remember the scale and recall their answers so the correlation will be inflated 2- some phenomenon change over time and therefore the test-retest approach is inappropriate for assessing consistency of a measure -often problematic to recruit the same people twice

external validity

is the extent to which a test measures what it intends to measure. There are 4 aspects of external validity 1-concurrent validity 2-construct validity 3-content validity 4- face validity -the degree to which the findings can be generalized to other populations and settings -can the results be replicated with other operational definitions? with other participants? in other settings? -a study can be internally valid (have cause-and effect relationship) but not ___ (the effect is only true for the particular settings, procedures and participants. -non experimental designs are usually higher on construct validity than experimental designs

concurrent validity

is whether scores on a new measure positively correlate with other measures known to test the same construct. This process is called cross validation -groups that should theoretically differ on the measure show the predicted difference. i.e. GRE-people high vs low in IQ

Face validity

is whether the test items simply look like they measure the construct -subject judgement; on its face does it seem to be a good translation of the theoretical concept?

Interaction

may arise when considering the relationship among three or more variables, and describes a situation in which the simultaneous influence of two variables on the third is not additive. Most commonly, ___ are considered in the context of factor analysis -when the effect of one IV on the DV changes (depends on) the other IV -this is a higher order effect (more information than the main effect) i.e. between adding sugar to coffee and stirring the coffee. Neither of the two individual variables has much effect on sweetness but a combination of the two does.

Reliability

means how stable the measure is -Is the measured variable close to the true ability/characteristic that we wish to measure? -how far are we from the true score? -do we capture a real score? or just random variance? =consistency of a measure. if your measure is __, it will produce the same scores everytime it is used. if it is un___, it will generate a different score everytime it is used -the more reliable your measure is, the less error it captures =consistency and repeatability of a measure -generalizability of a measure

Validity

means how well the test measures a construct -Donald Campbell and Donald Fiske created the multitrait-mulimethod technique to determine the validity of tests

Randomness

means that in selective a sample, each member of the specific population has an equal chance of being selected- that no weight or preference will enhance the selection chances for some members and weaken those chances for others. i.e. public opinion polling relies heavily on the concept of ___. -people are randomly assigned to conditions -each person has the same chance to be either in the experimental or control group -therefore, apart from the IV there should be no systematic difference between the groups- if there is a difference (beyond the IV) it is due to chance (=error variance) -researchers can assume that the characteristics of the participants of the two groups are roughly equivalent (i.e. on dimensions such as income, motivation, IQ, SES, etc)

criterion-reference tests

measure mastery in a particular area or subject (the final exam of a course, i.e.)

Implicit associations test (IAT)

measures behavior by reaction time; a good measure for sexism -measures the association between two or more concepts in a persons mind -are people faster to associate women with career or women with family?

internal validity

measures the extent to which the different items within a measure hang together and test the same thing -the degree to which a cause and effect relationship can be established -refers to the research design. Ask yourself: based on this design, can i conclude a cause and effect? what else (external factor) might explain the results/ -more experimental control--> less alternative explanations--> stronger internal validity -measured my cronbachs alpha

Type I error

mistakenly rejecting a null hypothesis -occurs when you incorrectly reject the null hypothesis- that is, you thought your findings were significant but they were really only caused by chance

Correlational vs experimental Research

most empirical research belongs clearly to one of these two general categories. In ___1 research, we do not (or at least try not to) influence any variables but only to measure them and look for relations (___1) between some set of variables, such as blood pressure and cholesterol level. In ___2, we manipulate some variable and then measure the effects of this manipulation on other variables. i.e. a researcher might artificially increase blood pressure and then record cholesterol level. Data analysis in___2 also comes down to calculating correlations between variables, specifically those manipulated and those affected by the manipulation. However, ___2 data may potentially provide qualitatively better information: only ___2l data can conclusively demonstrate causal relations between variables. i.e. if we found that whenever we change variable A then variable B changes, then we can conclude that "A influences B". Data from ____1 can only be interpreted in causal terms based on some theories that we have, but ____1 data cannot conclusively prove causality.

fatigue effect

order effect where participants become tired, bored, and distracted from one measure to the next and therefore respond differently

contrast effect

order effect where participants react more strongly to the second measure because they compare it to the first

practice effect

order effect where people gain more experience and get better at the task (DV) and therefore respond differently

Descriptive statistics

organize data from a sample by showing it in a meaningful way. They do not allow conclusions to be drawn beyond the sample. The five most common forms of descriptive statistics are: 1. Percentiles 2. Frequency Distribution a.nominal variables b. Ordinal variables c. Interval variables d. Ratio variables 3- Graphs a. frequency polygon b. histogram c. bar graph 4. Measures of central tendency a.mean b.median c.mode 5- Variability a. Range b. Variance and SD

completely randomized design

probably the simplest of experimental design in terms of data analysis and convenience. With this design, subjects are randomly assigned to treatments. In this design, the experimenter randomly assigns subjects to one of two treatment conditions.

one tailed tests

puts the entire significance level on a designated side of the distribution

empirical/criterion keying approach

refers to an approach to test Development that emphasizes the selection of items that discriminate between normal individuals and members of different diagn ostic groups, regardless of whether the items appear theoretically relevant to the diagnoses of interest.

Statistical inference (inferential statistics)

refers to sampling statistics and the process through which inference is made to whole populations through sampling procedures. Such inference requires careful attention to the concept of randomness in samples -inferring characteristics of a population from a sample of data -allows you to generalize findings from sample to a population, which is a larger group from which the sample was drawn. Statistics here refers to numbers that describe a sample, and parameters refer to numbers that describe populations. We use statistics to estimate population parameters. That is, we use statistics to predict or estimate what happens outside the sample.

Variability

refers to the relationship among all the scores in a distribution. Are they clustered closely around the mean or are they widely scattered? The term variance and standard deviation are measures of this variability. the standard deviation, is in effect the square root of the variance. Interpretatively, if you were comparing two SDs (one being 3.7 and the other being 1.2) you would know that the scores in the second distribution are generally closer to the mean and less scattered than the scores in the first distribution. If you increase of decrease each term in a distribution by a fixed amount, the variance and the SD remain unchanged. In effect, you have not changed the scatter of the distribution around the mean. If each term in a distribution is multiplied by a constant, the original SD would be affected in the same manner as every other score in the distribution (the resulting SD being the original SD times the constant) Because a SD squared would be its corresponding variance, the effect upon variance of multiplying each score in a distribution by a constant would be that or multiplying the original variance by the square of that constant. i.e. if ea\ach score in a distribution were multiplied by 2, and the original variance had been 6, the resulting variance would be 24. Squaring the constant makes it 4, so 4x6=24. In case of division by a constant, division of the square of the constant would yield the resulting variance -provides additional information to the central tendency and tells us how scores are spread out overall. includes range, variance and SD

Nonequivalent control group

research problem when a problematic type of control group is used when an equivalent one cannot be isolated

Illusory corelation

research problem when a relationship is inferred when there is actually none. i.e. many people insist a relationship exists between physical and personality characteristics, despite evidence that no such relationship exists

Experimenter bias (Rosenthal effect)

research problem when researchers see what they want to see. This effect is minimized in a double-blind experiment

demand characteristic

research problem when subjects act in ways they think the experimenter wants or expects -any feature of the experiment that might inform participants of the purpose of the study, and therefore bias their responses -the assumption here is that the participant is motivated to cooperate with the demands of the study and might act in ways that would fir with the experimenters expectations -therefore, it is not the treatment that caused the effect but the attempts to fit with expectations Overcoming demand characteristics: 1. cover story to disguise the exact purpose of the study -participants should not be able to figure our the predictions 2. Use implicit measures -word completion 3. Add filter tasks and filler items to disguise the purpose of the study -add another session for testing the DV 4. Ask the participants specifically to be honest, and assure them that there is no right or wrong answer

Social desirability

research problem when subjects do and say what they think puts them in a favorable light (i.e. reporting they are not racist even if they really are)

Cohort effects

research problem when the effects that might result when a group is born and raised in a particular time period

Selective attrition

research problem when the subjects that drop out of an experiment are different from those that remain. The remaining sample is no longer random -the tendency for some people in psychological experiments to be more likely to drop out than others. This can threaten the validity of the experiment

Reactance

research problem when there is an attitude change in response to feeling that options are limited. i.e. when subjects react negatively to being in an experiment by intentionally behaving unnaturally or when an individual becomes set on a certain flavor of ice cream as soon as he is told it is sold out -occurs in response to threats to perceived behavioral freedoms. i.e. when an individual engages in a prohibited activity in order to deliberately taunt the authority who prohibits it.

Design considerations

researchers are generally interested in how an IV affects a DV in a population. Because it is usually impossible to include all members of a population in a study, a sample or subgroup is drawn from the population. To make inferences about a population from a sample, the sample must be representative of the population and unbiased. This is most likely achieved with random sampling. However, sometimes random sampling is not feasible, and people use convenience sampling (like students in an intro to psych course) instead. To make the results for generalizable, though, researchers may use stratified sampling

alpha level

researchers cannot always know for certain whether their findings are correct, but certain standards are accepted. Most researchers use a significance level, or ___, of <.05 or <.01. This means that the chance that seemingly significant errors are due to random variation rather than to true, systematic variance is less than 5/100 or less than 1/100, respectively

Negative correlation

simple and linear. As one variable goes up, the other goes down. i.e. exercise and weight. With other factors held constant, as the amount you exercise increase, the amount you weight decreases

Positive correlation

simple and linear. As one variable increases, so does the other. i.e. food intake and weight are positively correlated; the more you eat the more you weigh

One-way ANOVA

simply tests whether the means on one outcome or dependent variable (i.e. height or level of anxiety from an anxiety scale) are significantly different across groups

counterbalancing

solution to order effects: -half the participants receive the treatment in a certain order (male applicant than a female applicant) and the other half receives the opposite order (Female applicant and then a male applicant) -order is added to the design as another IV

Creating measures

statistics are an important part of creating new tests or measures. They ensure that the measures are on target. -Tests are standardized (or tried out on huge groups of people) in order to create norms -Criteron reference tests -domain-reference tests -Reliability -Validity

independent t test

t tests for two independent means- suppose a researcher wants to determine whether there is a significant difference between the IQ scores of 12 year old boys and those of 12 year old girls in a given school. The scores of the boys and those of the girls have been obtained independently and, in effect, the comparison is between two sample means. This kind of setting- a comparison of two independent means is perhaps the most common and more often used t-test procedure

dependent t test

t tests of related measures- there is a relationship between the measures being obtained. i.e. one might take the above group of 12 year old boys and give them an IQ test just before instituting an intensive educational program and then administer the IQ test again at the conclusion of the program. The t-test would be comparing two sets of two sets of measures obtained on the same people (before and after an experimental procedure was introduced) to determine whether a significant change had occurred in their IQ scores. In addition to using this procedure to test the same people twice, it is possible to use it to compare the performances of matched groups. In such groups, each member of group 1 has been matched with a specific member of group 2 in the critical dimensions (age, sex, background and so on). On the rare occasions when developmental psychologists have been able to assemble a large group of identical twins, it has been the norm to assign identical twin A to group 1, identical twin B to group 2. By following a similar procedure for each set of identical twins, the researchers could be sure of matched groups since, for each person in group 1, there was a person in group 2 with identical hereditary background. The researchers could now institute an experimental procedure to determine how much performance change was a function of the experimental procedure. Such comparisons could utilize the t test for related measures

within-subjects design

tests the same person at multiple time points and looks at changes within that person -in a ___, the same people participate in all conditions of the study -___ factor- the same participants are assigned to each level of the factor -AKA repeated measures designs because people are measured repeatedly advantages: 1. some research questions require a ___ 2. ___ are economical (less participants, less expensive) 3-the error item is typically smaller compared with between subjects (because it does not include individual differences) in this design, two threats to internal validity (our ability to inder that the treatment caused the outcome) do not exist 1. order effects 2. demand characteristics

ANCOVA (Analysis of Covariance)

tests whether at least two groups co-vary. Importantly, the __ can adjust for preexisting differences between groups

F-test

the __ distribution is formed by the ratio of two independent chi-square variables divided by their respective degrees of freedom Since __ is formed by chi square, many of the chi-square properties carry over to the __ distribution -___ values are all non negative -the distribution is non-symmetric -the mean is appx. 1 -there are two independent degrees of freedom; one for the numerator and one for the denominator -there are many different ___ distributions, one for each pair of degrees of freedom The __ is designed to test if two population variances are equal. It does this by comparing the ratio of two variances. So, if the variances are equal, the ratio of the variances will be 1.

observed frequency

the among of times an event actually occurred after a probability experiment or trial has been repeated a given number of times

mixed designs

the design involves at least 2 IV's. For one IV, the same participants are assigned to each level (within-subjects IV) and for the other IV different participants are assigned to each level (between-subjects IV)

goodness of fit

the extent to which the observed data match the values expected by theory. Assessed with chi square

experimental methods

the independent variable is directly manipulated (controlled by the researcher) i.e. manipulate different levels of exercise, measure anxiety A relationship is established when the IV affects the DV (they covary) i.e. more exercise less anxiety - the researcher controls the cause (IV) Two key advantages: 1. control of the order of occurrence (the cause IV always precedes the outcome DV) 2. Ruling out alternative (3rd variable) explanations by: a. controlling the study-setting b. randomly assigning people to conditions -a study adhering to ___ takes place in a controlled setting (often a lab). In order to draw causal conclusions from an experiment, the researcher must be able to control certain aspects of the environment

probability in relation to the normal distribution

the likelihood of a given score occurring

mode

the most frequently occurring score

Degrees of freedom (DF)

the number of frequencies free to vary for any given n (i.e. if there are three brands of soda being tested with a chi square design, __=2 (n-1)) -the number of values in the final calculation of a statistic that are free to vary

expected frequency

the number of times a specific outcome is expected to occur in a given number of repeats calculated by multiplying the events probability by the number of repeats. i.e. rolling a 6 on a number cube in 24 turns ___=1/6x24=4

cluster sampling

the population is divided int groups and them some groups are randomly selected to participate in the

Cardinal rule

the rule that the mean moves in the direction of the skew

statistical regression

the step beyond simple correlations; allows you to not only identify a relationship between two variables but also to make predictions about one variable based on another variable

variables

things we measure, control or manipulate in research.

Histogram

this graph consists of vertical bars in which the sides of the vertical bars touch. ___ are used for discrete variables that have clear boundaries and for interval variables in which there is some order. The bars are lined up in order -bar graph representing score intervals -interval is the characteristic

frequency polygon

this graph has plotted points connected by lines. These are often used to plot variables that are continuous (categories without clear boundaries) -lines connecting points on a graph -frequency is the characteristic

Bar graph

this graph is like a histogram except that the vertical bars do not touch. The various vertical columns are separated by spaces

Double blind experiment

those in which neither the subject not the experimenter knows whether the subject is assigned to the treatment or to the control group

Scientific approach

to the study of psychology involves: 1- a testable hypothesis 2- a reproducible experiment that can be replicated by other scientists 3- an operationalized definition (observable and measurable) of the concept under study

Hartley's F max test

used in the analysis of variance to verify that different groups have similar variance, an assumption needed for other statistical test the test involves computing the ratio of the largest group variance to the smallest group variance. The resulting ratio is Fmax, is then compared to a critical value from a table of the sampling distribution of Fmax. If the computed ratio is less than the critical value, the groups are assumed to have similar or equal variances. ___ assumes that data for each group are normally distributed, and that each group has an equal number of members

graphs

used to plot data. Includes frequency polygram, histogram and bar graph

Factorial analysis of variance (ANOVA)

used when an experiment involves more than one independent variable. The analysis can separate the effects of different levels of different variables. i.e. if you were studying the effect of brain lesions on problem solving, you have two independent variables (lesion and type of problem) and one dependent variable (Success with problem) Give each independent variable two levels apiece (with and without lesions, simple and complex tasks) This is a 2x2 design which would yield four different combinations for evaluation. A ___ analysis can isolate the main effects (the effect of lesions on problem solving and the effect of type of task on problem solving). More importantly, a ___ can identify interaction effects. Here, you can combine the IVs (do people with lesions do better on simple tasks than people without lesions do on complex tasks)

repeated measures design (treatment by subjects design)

uses the same subjects with every branch of research, including the control. i.e. repeated measurements are collected in a longitudinal study in which change over time is assessed.

statistics

values obtained from samples -a descriptive measure of a sample

variable scales

variables differ in how well they can be measured, i.e. in how much measurable information their measurement scale can provide. There is obviously some measurement error involved in every measurement., which determines the amount of information that we can obtain. Another factor that determines the amount of information that can be provided by a variable is its type of measurement scale Specifically, variables are classified into a-nominal, b-ordinal, c-interval or d-ratio

Ratio variables

very similar to interval variables; in addition to all the properties of interval variables they feature an identifiable absolute zero point, thus, they allow for statements such as x is two times more than y. Typical examples of ___ are measures of time and space. i.e. as the Kelvin temperature scale is a ___, not only can we say that a temperature of 200 degrees is higher than one of 100 degrees, we can correctly state that it is twice as high. Interval scales do not have the ___ property. - premised on the absolute zero. it can express twice and three times as much (i.e. weight, height) -these variables have order, equal intervals and a real zero-they can say it all. i.e. age- after an absolute zero of not being born, age increases in equal intervals of years

Statistical tests

when a researcher uses a ___, he or she is usually hoping to find that the sample statistics are significant. This means that the numbers that describe the sample (i.e. that men are taller on average than women, or that treatment groups in therapy trials perform better than control groups) are describing a real difference or pattern rather than just random variation. If findings are statistically significant, then researchers can generalize these same findings to the population

single-blind condition

when only the participant does not know which condition s/he is in

Skewed distribution

when the data points cluster more towards one side of the scale than the other, creating a curve that is not symmetrical. In other words, the right and left side of the distribution are shaped differently from each other. There are two types; positively skewed and negatively skewed

test of significance

when the researcher uses a __, he or she is hoping to reject the null hypothesis, which is the hypothesis that no real differences or patterns exist. If a test of significance shows that results were statistically significant (not likely caused by chance) then the null hypothesis is rejected

Content validity

whether the content of the test covers a good sample of the construct being measured (not just part of it)

construct validity

whether the test really taps the abstract concept being measured -the degree to which an operational definition of a variable represents the theoretical construct -refers to the variables in your study (their translation) -applies to the IV (in experiments, evaluate the manipulation --if you have manipulated something other than your IV, then you have a problem of ___ -applies to the DV (how adequately did you measure your construct) -refers to the validity of your manipulation and measures i.e. you developed 5 questions to assess self-esteem ('how good do you feel about yourself? how confident are you in your abilities?")- are these questions assessing your concept of self esteem?

Probability type I and Type II error

you have 10 balls in your backpack. Five are red, three are green, two are yellow. Your backpack is specially designed with a ball dispenser and when you shake the backpack only one ball can come through the dispenser at a time. Since half of the entire group is red, the probability of a red ball coming through is .5; of a green ball .3; and of a yellow ball .2. Assuming you return all the balls to the backpack each time before dispensing, these probabilities will continue unchanged. The normal distribution discussed earlier is a bit like the colored balls. Theyre just stacked up in a very normal, symmetrical way. The probability of a ball being within one SD of the mean is .68 (.34 on each side of the mean); between one and two SD of the mean .28 (.14 on each side of the mean); and beyond 2 SD from the mean .04 (.02 on each side of the mean). When scientists set a significance level they ask themselves "what is the chance probability of mistakenly rejecting the null hypothesis (Type I error) or what is the chance probability of mistakenly accepting a null hypothesis (Type II error). The scientists set their significance level based on the risk they are willing to take of making an erroneous rejection (Type I error) or acceptance (Type II error).

convergence validity

your measure correlates with other measures that assess the same/similar theoretical construct -a GRE test correlates positively and strongly with other intelligence measures -if no other scales exist, the correlation between the new scale and measures of related constructs is assessed

discriminant validity

your measure does not correlate with other measures that assess different theoretical constructs -a GRE does not correlate with a test of motor skills ** note that here, stronger validity is reflected in a weaker correlation

predictive validity

your measure predicts future behavior that should predict theoretically -a GRE test should predict grades in graduate programs

Longitudinal design

Particularly in developmental research, psychologists need to study people at different ages. ____ involves studying the same objects at different points in the lifespan and provides better, more valid results than most other methods. However, ___ are costly and require an enormous time commitment, therefore many times a cross-sectional design is used

Step 1: Set up the null hypothesis

____ means "no difference". your ____ would be that reading speed is not affected one way or the other by carrot eating. actually, you believe that carrot eating has an effect on reading speed, so this hypothesis is one you hope to disprove later in the procedure. You are hoping that the difference you will find will be sufficiently great that you can disprove (reject) the ____ expressed here. Note, however that rejecting the ____ does not mean that you can prove that carrot eating affects reading speed. A research hypothesis is never proven, per se.

negatively skewed distribution

a distribution is ___ if the scores fall toward the higher side of the scale and there are very few low scores. ___ data is also referred to as skewed to the left because that is the direction of the long tail end of the chart. in this type of distribution, the mean is usually less than the median. -scores are bunched at the right and tall off or skew to the left -Central tendency sequential occurrence in a ___ is mean, median and mode (in that order)

population

a large group of people such as women, college students, stockbrokers or depressed patients

field study

an experiment that takes place in a naturalistic setting. Field studies generally have much less control over the environment than laboratory experiments do. For this reason, the field generates more hypotheses than it is able to prove

Placebo

an inactive substance or condition disguised as a treatment substance or condition. It is used to form the control group

variance

another statistical expression of difference from the mean (its the standard deviation squared. tells us how much variation there is among n number of scores in a distribution. To calculate ___,: 1-you figure out how much each score differs (or deviates) from the mean by subtracting the mean from each score. 2-Then you must square each of these devation values (this gets rid of negative values that result when scores fall below the mean). 3- Now, you add all these squared deviations to get the sum of squares. 4-Now divide this sum by the number of scores you had in the first place, by n-this gives you the __ of the sample. But remember, all of these values were squared, so to find the average deviation or SD from the mean you take the square root of the variance.

parametric tests (statistical procedures)

are based on the assumption that the population from which a sample has been drawn is in a normal distribution

nonparametric tests

are not dependent on the assumption that the population from which the sample has been drawn is in a normal distribution. i.e. chi square test

Correlation

by its name, suggests a co-relation. It is used to determine whether there is any systematic relationship between two sets of measurements or observations. The ___ is used to describe such a relationship is expressed in a range from +1 to -1. - A zero would indicate no relationship - a +1 would indicate a perfect positive relationship - a -1 would indicate a perfect negative relationship -and a +-1.1 or above would indicate a computational error has been made. ___ never exceed 1.0. it is important to realize that the degree of correlation is expressed by the number itself and not by its sign, i.e. between the numbers +0.5 and -0.7, the greatest degree of correlation is -0.7. The sign merely indicates in what direction the relationship exists. -part of statistics which is neither purely descriptive or purely inferential. __ can only show relationships (not causality) between variables. There are four types: 1-positive 2- negative 3- curvilinear 4-zero 0.0-.03=none to weak 0.3-0.69-moderate 0.7-1=strong

standard error of the mean

calculates how "off" the mean might be in either direction -estimates the variability between sample means that you would obtain if you took multiple samples from the same population. estimates variability between samples while the SD measures variability within a single sample

Cohort-sequential design

combines longitudinal and cross- sectional approaches

Between-subjects design

compares 2 groups of people at the same time point.

Quasi-experimental design

compares 2 groups of people like an experiment, but this design is used when it is not feasible or ethical to use random assignment. i.e. it is not ethical to assign one group of people to smoke for 20 years. so, that have shown that smoking causes lung cancer are quasi-___, since there was no random assignment to determine if people were smokers or not - they suffer from threats to internal validity

median

constitutes a midpoint of the sample scores (half the distribution scores are higher than, and half are lower than) - to find the __of a set of numbers, first line the numbers in ascending order. Find the value that lies in the center of the row. If there is an even number of values in the set, take the average of the two middle values.

Distinction between descriptive and sampling statistics

descriptive-statistics approach requires a person to specify a given population of interest and then collect measurements from all the members of that population. You can begin to imagine the difficulty of accomplishing this kind of measurement collection when you think of populations such as democrats or republicans. The more typical situation would involve having measurement access to a smaller group selected from the larger population of interest. The smaller group is known as a sample, and the statistics used in analyzing data collected from the smaller group are known as sampling statistics. Anlayzing the data from the sample- assuming that a sample is representative of the population has been obtained- you can then make generalizations from the sample to the population.

Frequency Distributions

explain how the data in a study looked. The distributions might show how often different variables appeared. It is the scatter pattern of scores

Ethical concerns

extremely important in research, and ethical standards in psychological research are very high. All studies are approved by an Institutional Review Board (IRB), which makes sure that the experiments are safe for participants and do not cause any unnecessary harm. Before participating in a study, all subjects must be provided with information about the risks and benefits of being in the study and then sign a consent form indicating that they are aware of the risks involves. However, these strict ethical guidelines did not always exist. The Milgram experiment was later deemed unethical because of the lasting psychological distress experienced by the participants. The Milgram experiment was the catalyst for higher ethical standards in psychological research

Cross-sectional design

in which different subjects of different ages are compared. This is faster and easier than the longitudinal design

Type II error

mistakenly accepting a null hypothesis -are when you wrongly accept the null hypothesis, in other words tests showed your findings to be insignificant when in fact they were significant

Curvilinear correlation

not simple and not linear. It looks like a curved line. i.e. arousal and performance- low arousal and high arousal lead to poor performance, but a medium amount of arousal leads to successful performance

random assignment

occurs when people are assigned randomly to either experimental or control conditions -you can have a study which involved random sampling but did not involve random assignment to conditions, and the opposite

Random sampling

occurs when people are drawn randomly from a population -you can have a study which involved random sampling but did not involve random assignment to conditions, and the opposite -with this sampling procedure, every member of the population has an equal chance of being chosen for the sample

research design

refers to how a researcher attempts to examine a hypothesis. Different questions call for different approaches and some approaches are more scientific than others

Acquiescence

research problem when people agree with opposing statements

Hawthorne effect

research problem when subjects alter their behavior because they are being observed. This also applies to workers altering their behavior for the same reason

Placebo effect

research problem when subjects behave differently just because they think they have received the treatment substance or condition

within subject design

tests the same person at multiple time points and looks at changes within that person

predictive value

the degree to which an independent variable can predict a dependent variable

Generalizability

the degree to which the results from an experiment can be applied to the population and the real world

Statistics

the process of representing or analyzing numerical data

Zero correlation

there is no relationship between the variables