Final Crinq 1
interpret the following: -intention to treat analysis -power analysis -Type 1 and Type 2 error
*-Intention-to-treat analysis:* means all patients who were enrolled and randomly allocated to treatment are included in the analysis and are analysed in the groups to which they were randomized. i.e. "once randomized, always analyzed" ****should conduct intention to treat analysis when there is a high attrition rate *-Power Analysis (1-B):* --ability to find a statistical significant difference when a real difference exists. (Kind of like the concept of sensitivity: ability to find a disease in those who actually have the disease) --A power analysis is commonly conducted to determine how many subjects in a research study is needed to find a statistical difference. Probability that a test will lead to the rejection of the null. Determined by: 1) alpha level 2) sample size (larger N = larger power) 3) effect size -The probability of NOT making a type 2 error (1-B). When there is low power (small sample size), a type 2 error is usually committed -"B" = chance of making a type 2 error *Type 1 error:* said there was a difference when there actually wasn't *Type 2 error:* said there is no difference when there actually was a difference
Be able to describe the research scenarios -Systematic Review -Meta-analysis -RCT -quasi-experimental trial (either no control and/or randomization) -case series -case study -single system design
*-Systematic Review:*-collects and critically analyzes multiple research studies or papers. Aims to provide a complete, exhaustive summary of current literature relevant to a research question. Strongest form of literature *-Meta-analysis:* -A systematic overview of studies that *pools results* of two or more studies to obtain an overall answer to a question or interest. *Difference between systematic review and meta analysis:* A systematic review answers a defined research question by collecting and summarizing all empirical evidence that fits pre-specified eligibility criteria. A meta-analysis is the use of statistical methods to summarize the results of these studies. *-RCT:* prospective and experimental (IV is manipulated and DV is collected under controlled conditions). *A group of patients is randomized into an experimental group and a control group.* These groups are followed up for the variables / outcomes of interest. *-quasi-experimental trial:* lacks control and/or randomization. Can be from within subject, between subjects, or in a single subject -*case series:* describes characteristics of a group of patients with a particular disease or who have undergone a particular procedure. Design may be prospective or retrospective. No control group is used in the study, although the discussion may compare the results to other published outcomes. ex) A group of physical therapists documented the outcomes of an eight-week general aerobic, flexibility and strengthening program in 10 persons with chronic low back pain over a span of a year • Several case reports put together to illustrate a single point • N > 1 but usually < 20 • High potential for bias, but valuable "stories from the field" *-case study:* description of individual response to treatment- nonexperimental. Its more like a story of the person. N=1 *single system design*- -We are comparing the performance of two different groups on a single theme (has a low # of subjects) *CAN be one person too if time is emphasized* (e.g. randomly assign clients to experimental group where they receive CBT, or a control group where they do not, and assessing depression levels of both after the intervention period). -some clients receive intervention and some clients do not -involve repeated measurements of a behavioral response over time -difference between case series, case study, and SSD: case series has no control group, SSD does, case study is non-experimental, SDD is, case series has more than one subject, case study has only one subject. -single subject design: able to draw conclusions about the effects of treatment of a single subject under CONTROLLED conditions - *experimental*
Differentiate between: -construct/content -criterion -concurrent validity
*Construct:* -the degree to which a instrument/test measures what it claims, or purports, to measure. (ex. health questionnaire, surveys)-- Do the Qs really evaluate health?) *Criterion:* -instrument can be used as a substitute for an established gold standard. -To measure the criterion validity of a test, researchers must calibrate it against a known standard or against itself ex) A physical therapist would like to compare the results he obtains from a new isokinetic testing equipment to the Biodex (which is considered to be the gold standard in isokinetic testing). *Concurrent:* -the extent to which the results of a particular test, assessment or measurement correspond to those of a previously established measurement for the same construct. NOT COMPARED TO A GOLD STANDARD ex) one IQ test compared to another IQ test you are comparing A against B when A and B are similar and B is already currently being used as a test
descriptive and inferential statistics
*Descriptive* -formulate a description of reported data; results from a population -sample characteristics or results of the study e.g % of people ... mean, median, mode (measures of central tendency) and range, standard deviation, variance (measures of variability) *Inferential* -used to make comparisons -determine the effects of different factors on an outcome -determines if there are differences between groups -make it possible to describe a relationship between the IV and DV -based on a normal distribution ex) p values are used: P stands for probability
clinical vs. inferential statistical measures
*Inferential statistics*: used to make comparisons and determines the effects of different factors on the outcomes. Makes it possible to describe a relationship between the IV and DV Inferential statistical measures: P-value (<0.05 = groups are statistically significant different). Chi Square, Mann Whitney U, Kurskal Wallis, t test, ANOVA, *Clinically relevant statistics*: evaluate the importance of outcomes for patient care Clinical statistical measures: 1) MCID, MCD, NNT, CIs, Effect Size, Power Analysis, Sensitivity and Specificity, mean, range, SD Example of Effect Size: Standardized difference between 2 means. Refers to the magnitude of the impact of some variable on another. A meta-analysis or systematic review compares studies to each other to find differences between 2 means. Cohen's D is an example: 0.8 or greater is what we want it to be Example of NNT: NNT=4 then investigators need to treat four persons in order for one additional person to achieve a predetermined clinical improvement Example of tests of diagnostic accuracy: what would the research classification be if researchers wanted to evaluate the ability of the new "exam" in evaluating PCL tears. They compare their findings to an MRI considered to be a gold standard
Statistics of Diagnostic Accuracy (know how to interpret) aka. know contingency table
*Sensitivity:* ability to obtain a positive test when the testing condition is truly positive ex) The sensory organization test is a valid measure in identifying those with vestibular dysfunction 15% of the time amongst those who have the condition *Specificity:* ability to obtain a negative test when the testing condition is truly negative ex) The sensory organization test is a valid measure in identifying those without vestibular dysfunction 15% of the time amongst those without the condition *PPV:* portion of those with a positive test who have the disease *NPV:* portion of those with a negative test who do not have the disease *positive likelihood ratio:* the BEST indicator of a clinical test in identifying a patient/client with a condition when compared to those without the condition *negative likelihood ratio:* the test appropriately identifies those without the condition when compared to those with the condition
Within a 2 way ANOVA, what are the main effects and the interaction effect
*main effect*: "collasping" or taking the average across the rows or the columns -taking an average of the 3 different degrees of hip flexion (under 3 difference conditions)--> across column -take an average of the hip flexion and hip extension under that 1 condition--> across row *interaction effect:* one variable is dependent on the level of the second variable ex) do I get a greater degree of hip flexion (1 IV) when I do quick or prolonged stretching (another IV) prior to testing for hip flexion. Same can be done for hip extension.
what does p= 0.05 represent vs. an alpha value of 0.05
*p-value:* probability that an observed difference occurred by chance (determined by statistical test). A small p-value (typically ≤ 0.05) indicates strong evidence against the null hypothesis, so you reject the null hypothesis. pvalue of 0.05 means there is a 5 % chance that the difference between groups was due to chance. There is a 5% chance of committing a Type 1 error *alpha level:* the level that the significant differences will occur due to chance ALONE. So 0.05 means that 5% of the time, the differences will be due to chance ALONE. Decreasing alpha value (drug studies) decreases the chance for Type 1 error to occur. Alpha value sets the standard for how extreme the data must be before we can reject the null hypothesis. The p-value indicates how extreme the data are. We compare the p-value with the alpha to determine whether the observed data are statistically significantly different from the null hypothesis
parametric vs. non-parametric
*parametric:* -refers to assumptions about the population from which the data was obtained. -random selection = normal distribution= continuous data -more robust when we find differences; they are more credulous *nonparametric:* -no inferences are generalized to the population of subjects -useful for small homogenous sample sizes -easier to find a difference but not very credulous
Statistical vs. Clinical significance
*statistical significance:* -tells how sure you are that a difference or relationship exists -is expressed in terms of a probability (p< 0.05): The probability of observing an effect given that the null hypothesis is true *clinical significance:* -evaluates the importance of outcomes for patient care -provides information about odds ratio, effect size, and clinical relevance (sensitivity and specificity) -compares numbers from one study to the next -assesses between group differences
True statements
-p value assesses the findings due to chance alone -a p value of .01 is not more statistically significant than a p value of .04 -a Pearson or Spearman correlation are not robust tests in the evaluation of reliability. Remember, it's only correlation
What do limits and Booleen operators do
-the use of "AND" or "OR" are termed Booleen operators. -If a therapist were looking for a journal article and would like to limit their search including two search terms, they would most appropriately use Booleen operator "AND" -using the term "OR" will expand your search when compared to using "AND" -MeSH headings are useful to find alternative synonyms for your search -"limits" will narrow your results down
2nd graph of CIs says that there are multiple studies present with a control and hip OA group. CIs are plotted with the axis being effect sizes which means that these are differences between the means (can make prediction about statistical significance)
-this was taken from a meta-analysis. The CIs that cross the zero are not statistically significant results. -When looking at the pooled results, we can see that the control group is faster than the hip OA group because the walking speeds are less = faster
Within an ANOVA, what is the purpose of post-hoc testing
-used when authors reject the null hypothesis -the purpose of running this is to determine where the differences lie because without running it, we know there are differences if the null is rejected but we don't know where they lie. Helps to uncover the differences between 2 groups
Differentiate between when each should be used: give an example 1) t-test 2) 1-way ANOVA 3) 2-way ANOVA
1) compares means of 2 levels of groups on 1 IV ("differences") -not used for 3 or more levels because the % chance of committing a type 1 error increases, so that's why we use an ANOVA instead. If you do use 3 t-tests, then you increase your chances for a type 1 error. A Bonferonni statistic should be incorporated to decrease the chance of a committing a Type I error. ex) you would like to evaluate the differences between two groups (based on gender, occupation, and age) on their ROM. One group receives stretching and the other receives manual therapy: (Paired t-test = dependent t-test) ex)Scenario #2 A PT was interested in evaluating if person taking arthritis medication and participating in an aquatic therapy program were more independent than those taking medication alone. The dependent measure time on "Get up and go" test: (groups are independent of each other and the data is continuous = parametric so the test used is Independent t-test) ex) Scenario #7 Two PT back specialists would like to evaluate EMG activity (measured in microvolts) using fine wire electrodes of the transverse abdominus. Their two scenerios will be with the use of a backbrace and without. What statistical test would be appropriate? (Repeated measures t-test: have everyone do the backbrace and everyone without) 2) compares 3 levels of groups on 1 IV against 1 DV ex) what is the degree improvement of hip flexion when subjects do prolonged stretching, quick stretching, and no stretching (control) 3) compares 2 IVs on 1 DV. In a two-way ANOVA, not only can the main effect of each IV be calculated, but so can any interaction effect between the different IVs. (i.e pre and post to PNF, ballistic, and control). (i.e male and female results to PNF, ballistic, and control) If question was looking for change of ROM instead of pre and post, then it would turn out to be a 1 way ANOVA. ex) You would like to evaluate the difference between two matched groups (based on gender, occupation, and age) on their CROM. They are measured at baseline, 6 months, and 12 months. (repeated measures 2 way ANOVA)
Scenario #1 Two raters evaluate 25 elderly home residents. They are analyzing the *similarity* of their abilities to categorize individuals into their level of assistance. The 3 categories are independent, stand by assist (SBA), and dependent.
Purpose: similarity of their ability = reliability = ICC or KAPPA IV: raters DV: lvl of assistance (ordinal data = non parametric) TEST: Use Kappa statistic
Scenario #7 Two PT back specialists would like to *evaluate* EMG activity (measured in microvolts) using fine wire electrodes of the transverse abdominus. Their two scenerios will be with the use of a backbrace and without. What statistical test would be appropriate?
Purpose: test for differences IV: intervention (backbrace or without = 2 lvls) DV: EMG activity in microvolts which is continuous data = parametric participants run through each scenario then it's dependent TEST: repeated measures t test
Scenario #9 You would like to determine the *difference* in subjects perception of "coolness" from treatment of ice massage, ice pack and cold bath. The "coolness" is determined on five ratings from 1-5.
Purpose: test for differences IV: treatment DV: perception of "coolness" ( ratings of coolness = ordinal = non parametric. 3 levels of groups: ice massage, ice pack, and cold bath TEST: Kruskal Wallis ANOVA
Scenario #5 A researcher is interested in the *effect* of body position (supine and sitting) on a person's ability to relax. 22 subjects are randomly assigned to supine or sitting conditions first to minimize a learning effect. EMG is measured via biofeedback in microvolts. The microvolt levels are put into a rank (low, low plus, medium etc. to high)
Purpose: test of differences IV: body position (supine and sitting = 2 levels) DV: relaxation (EMG microvolts = low medium high = ordinal = non parametric) supine and sitting conditions are independent of each other TEST: Mann-Whitney
Scenario #2 A PT was interested in *evaluating* if person taking arthritis medication and participating in an aquatic therapy program were more independent than those taking medication alone. The dependent measure time on "Get up and go" test
Purpose: test of differences IV: intervention (aquatic therapy vs medication alone = 2 levels) DV: time (ratio data = parametric) arthritis medication and aquatic therapy are independent of each other!! TEST: Independent T-test
Scenario #6 A researcher is interested in evaluating the *effects* of the PNF stretching and ballistic stretching on straight let raise (SLR). There are 3 groups: PNF, ballistic, control. This is a single blind study where the evaluators are not aware of the treatment. They pursue a pre and post test to evaluate before and after. The statistical test will compare the overall difference between pre and post by averaging group means as well as any differences between the 3 groups.
Purpose: test of differences IV: type of stretching (PNF vs ballistic) 3 groups = 3 lvls DV: SLR (pre and post test) Stat test will average 3 group means (pre and post) = goniometric reading = ratio data = parametric) 2 DVs: SLR pre test AND SLR post test TEST: 2 way ANOVA because researchers want to compare pre AND post test data
Scenario #4 You would like to determine a *relationship* between females who are at low, medium, and high risk of falling based on a clinical outcome measure (rated on a scale from 0,1,2,3,4,5 low to high) and the number of falls in a week.
Purpose: test of relationships IV: none DV: risk of falling and # of falls (low, medium, and high = ordinal data; # of falls = ratio data.... SO you *use non-parametric when one is ordinal and one is ratio* TEST: Spearman
So a CI graph that shows multiple studies as well as sensitivity and specificity data is what (graph 1 in her final review slides) (DOES NOT TALK ABOUT EFFECT SIZES= no differences between means!
RCTs Systematic Reviews The sensitivity is variable which means that it's bad overall when looking at all of the studies together as one (range is 0.2-1)-- yikes Specificity of the one study shown is really good! Shows that this study is good at finding those that don't have the condition in those that truly don't have the condition. In other words, the study is useful in identifying those that don't have the condition.
Search Engines to use for finding peer reviewed research
•Conduct a literature review using electronic database search engines: *SOLAR*- Searches all of CSS's library resources at once. Articles are not all peer reviewed. *PubMed*- Contains ONLY peer reviewed articles. *CINAHL*-Comprehensive coverage of journal literature in nursing and allied health. Titles in PT include Physical Therapy Reviews and Physiotherapy Theory & Practice. *Medline*- The premier medical database. It includes indexing in Medline and full-text for important titles in Physical Therapy including Physiotherapy International and Physical Therapy. *APTA's: Hooked on Evidence*- only available to APTA members *Cochrane*- a collection of systematic reviews of evidence based medicine. *PEDRO* (Physiotherapy Evidence Database)-This database gives "rapid access to bibliographic details and abstracts. Does NOT include clinical practice guidelines, systematic reviews, or other clinical studies. Covers only intervention related literature. Also has a documented rating system for determining the quality of clinical trials (pedro scale). Rated a 10/10 is considered to have the least risk for bias *Google Scholar*- valid search engine. Searches the internet for journal articles. Searches all entries in pubmed as well. *****Mayo Clinical and Web MD ARE NOT REPUTABLE SEARCH ENGINES WITH PEER REVIEWED RESEARCH ARTICLES
Scenario #8 A student group of PTs were concerned about their 2 types of PT: instructor or cliniciam and wanted to compare *incidence* of a repetitive overuse problem between their instructors compared to a matched population of PTs working in a clinic.
Purpose: Proportion, Frequency, Incidence IV: type of PT (instructor or clinician) DV: incidence of injuries TEST: Chi Squared
What if you had the same scenario, but the researchers were wanting to determine the inter-rater *reliability* of the get up and go test (continuous =parametric) , which utilizes time as the outcome measure?
Purpose: inter-rating reliability = reliability = ICC or KAPPA IV: raters DV: time (ratio data = parametric) ICC: intraclass correlation coefficient
A researcher would like to determine the *relationship* between ROM of the knee and gait velocity.
"relationship" tells us it's either spearman or pearson. ROM and gait velocity are continuous data = parametric So test to use = Pearson Correlation
Purpose of stratified block randomization
*-are constructed to reduce noise or variance in the data* -a method to ensure that 2 groups are homogenous based on certain characteristics. -randomly selecting individuals based on identified characteristics from a population
standard deviation
-average distance from the mean -gives the PT a hint of how variable the results are -under 1 is good -increased SD means there is more deviation from the mean for that group
Confounding variables
-nonrandomization of subjects and variables related to subject design -instrumentation: make sure it's valid and reliable -blinding subjects or researchers -Hawthorne effect: occurs when a subject knows that he/she is participating in a research study and thus that affects how he/she performs in the research. Prior knowledge of being in a study affects the outcome. Randomize to countereffect it -sampling bias- control for this via randomization -repeated measures- control for this via randomization -attrition: occurs when subjects drop out of the study.. Control for this with large sample sizes
Define probability in your own words
-the extent to which something is probable -the likelihood that an observed difference occurred by chance
In the same study, the researchers want to determine if there is a significant difference in the number of falls in the 3 groups over a 2 year period. They also felt it was important to analyze the males and females in their group. What type of statistical test would they employ now? What are the null hypothesis in this study (note 3)?
1 null: training does not have any impact on balance 2 null: there is no difference in the # of falls in the 3 groups over 2 yrs. 3 null: there is no difference between the results from the males and females test to use: over 2 year period = NOT repeated measures. males and females now makes it a 2 way ANOVA SO...2 by 3 way ANOVA What type of statistic evaluation would detect if males perform differently DEPENDENT on females? --> interaction effect!! Remember... a 1 way ANOVA can show a main effect while a 2 way ANOVA can show a main effect and a interaction effect!
A researcher evaluates the ability of two different treadmills in the ability to "unweight" (i.e. Ultra G, Lightspeed and a control that does not unweight). The *participants run on each treadmill* wearing PEDAR insoles that evaluate the forces occcurring at the foot. The most appropriate statistical test to evaluate the change in forces should be:
1 way repeated measures ANOVA
Authors would like to compare three different forms of manual therapy (manipulation, mobilization, massage) *and also include a control group (that makes 4)* in persons with cervical immobility secondary to chronic whiplash. Their outcome measure is cervical range of motion. They measure CROM at baseline, 6 weeks and a 6 month follow-up *(that makes 3)*. Which of the following statistical analysis would be most appropriate in analyzing this data?
3 by 4 ANOVA
The same authors reported that the Crank test had a sensitivity of 91% with a confidence interval of 76-97%, which of the following is the MOST accurate interpretation of these findings?
95% of the time sensitivity of the CRANK test will fall between 76%-97% -significance is in relation to the differences between the groups so based on the information given, you can't make the judgment on whether or not these findings are statistically significant because the CI does not include differences. Effect size includes difference between two means
A researcher would like to evaluate the differences between tai-chi training, standard balance training and standard edcuation between two groups of elderly individuals: nursing home residents and community dwelling elders. The outcome variable is the number of falls sustained in a 2-year period. The MOST appropriate statistical test for this scenario should be
A 2 by 3 ANOVA Nursing home residents is one column and community dwelling elders is another column. 3 rows include: tai chi training, standard balance training, and standard education There are 2 IV's: type of balance training and elderly residence DV is # of falls, which is parametric
Difference between a 1 way and 2 way ANOVA
A one-way ANOVA has one independent variable while a 2-way has 2 independent variables
If an alpha was set at .05 and if authors obtain a p value for a comparison of .04 and another comparison was .01, this should most appropriately interpreted as:
Both were statistically significant; one had 4/100 chance of differences being due to chance while the other had 1/100 chance
independent variable vs. dependent variable
IV: presumed to be the cause of the dependent variable; the variable that is manipulated or controlled by the researcher DV: response or effect variable from the independent variable
You are analyzing the effect of balance training on geriatric adults. One group receives standardized balance training, another engages in Tai Chi and another does not receive any treatment at all. Balance is based on risk of falling: low risk, medium risk, high risk (based on the number of falls they have per day). Only post-test values are obtained.
IV: type of balance training DV: degree of balance 3 levels of 1 IV AND its nonparametric because (low, medium, high risk are ordinal data= rank) test to use: Kruskall wallis 1 way ANOVA BUT if it was pre and post test values were obtained, then it switches to a 2 way ANOVA
Which of the following is NOT an available method in limiting an electronic search? Year of publication Booleen operators Mesh Headings Type of research Language
Mesh Headings ; because all mesh headings does is find synonomous terms
Scenario #3 You would like to see if a *relationship* exists between romberg stance times and the number of falls (continuous data) in a year amoungst community dwelling female elders
Purpose: test of relationships IV: none DV: stance time and # of falls (ratio data =parametric) TEST: Pearson Correlation
Diagnostic Accuracy of clinical exams in detecting hamstring tears: A meta-analysis Intervention Reliability Systematic Review Prognosis Qualitative
Systematic Review
A researcher finds that the O'Briens test for detecting SLAP lesions has a 66% sensitivity when compared to arthroscopy. The MOST appropriate interpretation of these findings is:
The O'Briens test accurately detects SLAP lesions 66% of the time in persons who truly have SLAP lesions
If an exam reveals a +LR of 10.1, which of the following most accurately represents the findings:
There is a high liklihood that the patient will have the disease if tested positive compared to those without the disease
If a researcher fails to recruit an adequate number of subjects and does not find statistical significance (although a significant difference exists); this is termed a:
Type 2 Error
If authors had utilized pre/post continuous data, but found their data did not fall under the standards for normality (i.e. bell-shaped curve), they would then use a _____ statistical test to determine differences within a group
Wilcoxin
a) correlation (r value) can do what but cant do what b) what can't it evaluate
a) can determine statistical significance but CANT PREDICT A RELATIONSHIP b) a pearson correlation is bad at evaluating a parametric measure and is bad a measuring reliability
a) the exam is useful in identifying those without the disease A) the exam is useful in identifying those with the disease b) True positive B) True negative c) The test is good at identifying those with disease amongst those who have the disease C) The test is good at identifying those without the disease amongst those who do not have the disease d) The test is good at identifying a *true negative rate amongst those who test negative* OR portion of those that tested negative who do not have the disease D) portion of those that tested positive who have the disease OR the test is good at identifying a *true positive rate*amongst those who test postive
a) specificity A) sensitivity b) sensitivity or PPV B) specificity or NPV c) sensitivity C) specificity d) NPV D) PPV
hierarchy of best sources of evidence for searchable clinical questions (Oxford)
basic science: (narrative reviews, expert opinion, textbooks)- *WORST* small studies: case studies, case series, case control studies cohort studies large RCTs- *GOLD STANDARD* systemic reviews--*strongest form of evidence*
methods of sampling
cluster: high incidence of subjects in 1 area snowball: tell one person, they tell someone else, repeated convenience: easy to get from a place simple random: select lets say 50 ppl from a group of 300 ppl
Authors conduct a study to determine the effectiveness of the GRASTON tool-implemented massage technique on Achilles tendinopathy. They randomly assign 30 subjects into two groups: traditional therapy and traditional therapy plus GRASTON. Their outcome measure is the amount of pain change (baseline minus 6 week pain) on a 10 cm line. The most appropriate statistical analysis to evaluate this data is:
continuous data: 10 cm line--> parametric SO a t-test would be used 1 IV with 2 levels of groups plus parametric data = t-test
Determination of the upper limb tension test in detecting CTS when compared to nerve conduction Intervention Reliability or Validity Diagnostic Accuracy Systematic Review Prognosis Qualitative
diagnostic accuracy
given a case scenario, classify research into either: -intervention -diagnostic accuracy -prognosis -outcome measures (includes reliability)
diagnostic accuracy: what would the research classification be if researchers wanted to evaluate the ability of the new "exam" in evaluating PCL tears. They compare their findings to an MRI considered to be a gold standard outcome measures: what would the research classification be if a therapist would like to know the *reliability of using standardized questionnaires* in ppl with chronic low back pain. overall satisfication. (e.g. questionnaire pre + post experiment) prognosis: deal with correlation or level of improvement that may be attained. (predict): a prediction of something. GPA and GRE are predictors for the % pass rate of the NPTE for the 1st time intervention: tests for differences between something. (experimental) ex) A physical therapist would like to pursue a research study *comparing the effects of manual therapy, "therapy hands" (sham) and a mobility only group on shoulder internal rotation* in persons with shoulder impingement presenting with 40 degrees or less of internal rotation. Another physical therapist is blinded to the subject treatment allocation and performs all of the range of motion measurements (they take the average of three repeated measurements pre and post intervention). The researchers recruit 50 subjects in their outpatient orthopedic clinic where the subjects are randomized to treatment allocation. They evaluate the range of motion improvement (one score) from pre to post test
What are a few sound methods for conducting an experimental trial
experimental and control group large sample size (random selection) controlled manipulation of the IV
Should authors increase their alpha value from .01 to.05, they have:
increased their chance for type 1 error to occur (recall: in drug studies they decrease there alpha value to decrease the chance for type 1 error)
Effects of massage in marathon runners in decreasing DOMS: A randomized controlled trial Intervention Reliability or Validity Diagnostic Accuracy Systematic Review Prognosis Qualitative
intervention
3 measures of central tendency
mean (average) mode (most) median (middle when in order)
null vs. alternative hypothesis
null states there are no differences between population means and that any observed differences between groups are due to chance alternative states: contrary to the null: observed differences between two population means are not due to chance
Repeated measures design means
one independent variable was assessed repeatedly on the same subject over time or condition
If authors set an alpha value at .05, conduct a one-way ANOVA with a resulting p value of .25, the authors should then:
p value is greater than alpha value so accept the null
A researcher is investigating the effects of a Smith and a Norkin ASO on ankle proprioception in basketball players in the subacute stage of Grade II ankle sprain. They evaluate roprioception is measured using a force platform measuring sway (in cm).
purpose: test for differences (bc it says "effects") measuring sway in cm = continous = parametric data 1IV: 2 lvls of groups that are independent (dont relate in any way) of eachother test to use: independent t-test
3 measures of variability
range- max-min SD- average distance from the mean variance-the amount of time that the 2 groups share their variability
If the impingement exam is found to have a high rate of false positives, what will this affect: the sensitivity, or specificity?
specificity
If the Apley's exam is 70% sensitive and 50% specific for detecting meniscal tears when compared to a gold standard, what does this mean in your own words?
the test is moderate at finding the meniscal tear in those that have a meniscal tear. (70% of the time) the test is bad at finding people that don't have a meniscal tear amongst those that don't have a tear because it's only accurate half of the time
Authors found that a valgus measurement of the rearfoot was significantly associated with body weight (Pearson's r=.3; p=.02 with alpha set at .05). The most appropriate interpretation of these findings is:
there is a weak yet significant relationship between rearfoot valgus and weight r ranges from +1 to -1
A physical therapist would like to *determine the relationship* between the number of miles run and the number of chronic injuries sustained in marathon runners over a 10-year time period. The MOST appropriate statistical analysis should be:
time = ratio = parametric data Relationship tells us it's either Spearman or Pearson but since it's parametric, the answer is PEARSON CORRELATION
