Research Methods for SW: Ch.9
assign a replacement value to missing data: if you're creating an index out of several items, can use proportions based on what is observed
a way to deal with the problem of missing data; clue: averaging answers
construct the scale or index using alternative methods and see how each affects findings
a way to deal with the problem of missing data; clue: the safest and best method
stylistic differences: quantitative approaches _ use same wording and sequence of questions for all respondents - qualitative approaches _ use them
always, rarely
As a general rule, whenever the word _ appears in a question or questionnaire statement, you should check whether you are asking a double-barreled question
and
double-barreled question
asking for a single answer to a question that really contains multiple questions
double-barreled questions
asking for a single answer to a question that really contains multiple questions
one of the most common ways social work researchers operationalize their variables is by ....
asking people questions as a way to get data for analysis and interpretation
*The first step in constructing a scale is item selection. Items should have what?
*proper face validity* (your self-esteem scale should include items that only reflect self-esteem, not items that reflect depression - also known as unidimensionality in scale construction) , *adequate variance* (if everyone gives the same response for an item then that item won't have any use in constructing a scale) , *reliability and validity*
quantitative and qualitative approaches: similarities always shared in measurement principles are (p238chart)
-both try to use info resp. can understand -avoid dbl barreled questions: 1 ? at a time -relevant questions that resp. are capable of answering -avoid biased terms and items
list the advantages of matrix questions
-can use space efficiently, can provide quicker answers -may increase the comparability of responses given to different questions for the respondent as as for the researcher (bc can help answer other questions by being able to quickly review strength of answer on previous questions)
problems of matrix questions can be reduced by...
-interspersing positively and negatively worded statements to represent different orientations -making all statements short and clear
list the drawbacks of matrix questions
-items may be force into response categories when others are more appropriate -response set bias is an issue (may develop a pattern of response) -might assume all same direction (respondents might assume they represent the same orientation and may answer them incorrectly)
To guard against error, you will want to pretest the questionnaire in a dry run. What are some guidelines to follow?
1.small sample: 10 people or less 2.should be like the people you intend to include in study 3.NOT have to be a randomly selected sample 4.pretest participants should NOT participate in actual study 5.better to ask them to complete rather than look for errors
*likert scale
A composite measure of an attribute involving the summation of scores on a set of items that respondents typically rate their degree of agreement or disagreement. A question format frequently used in survey questionnaires. Respondents indicate their choice from the unambiguously ordered response categories: strongly agree, agree, disagree, strongly disagree
the Likert method is based on this assumption...
An overall score based on responses to the many items that reflect a particular variable under consideration provides a reasonably good measure of the variable.
Suppose the scale has too many positively worded items. you could change one positively worded item such as, "engaging in the EBP process will improve one's practice," to "...will not improve one's practice". How could you avoid using the word "not" by rewording the question?
Engaging in the EBP process makes practice too mechanistic - or - Engaging in the EBP process hinders the practitioner-client relationship.
One way to approach the problem of question ordering is to randomize the order.
False
True or False: If we find that our measurement instruments are reliable and valid when tested with one culture, then we can assume that they will be reliable and valid when used with other cultures.
False
True or False: A study can not use both qualitative and quantitative approaches
False *they need not be seen as mutually exclusive or in conflict
Complementary Functions: quantitative approach vs. qualitative approach
Generalizability vs. in-depth, theoretical understanding
Finish this paragraph with the sentence that fits best: Despite the desirability of mutually exclusive answer categories in closed-ended questions, more than one answer may often apply for respondents.
If you want a single answer then make this clear in the question
Is the following question double barreled? "If a friend of yours began to exhibit strange and erratic behavior, what do you think your response would be?"
No - it asks only about behavior that is both strange and erratic
What is the chief difference between quantitative and qualitative measure for asking people questions?
Quantitative measures are always highly structured, tend to use closed-ended questions primarily, and may be administered in either an interview or questionnaire format, whereas qualitative measures rely on interviews that are often unstructured and that mainly contain open-ended questions with in depth probes.
When attitudes are requested on a topic that few respondents have thought about or really care about, the results are not likely to be useful. This relates to which guideline for asking questions?
Questions should be relevant
The Gallup Organization has utilized a "secret ballot" format that simulates actual election conditions by giving the "voter" complete anonymity. This is an example of coping with which guideline to asking questions?
Respondents must be willing to answer
What is true of every questions and answer?
The meaning of someone's response to a question depends in large part on the wording of the question that was asked. : answer's meaning depends on question's wording
why do questionnaire's often have as many statements as questions?
The researcher is interested in determining the extent to which respondents hold a particular attitude or perspective. (so provide statement and ask if agree or disagree)
matrix questions
Save space and reading time when several items share the same set of response choices - a particular kind of closed-ended question used to ask people about their attitudes , strongly agree, agree, disagree, ect....
The respondent should be able to read an item quickly, understand its intent, and select or provide an answer w/out difficulty. You should assume that respondents will read items quickly and give quick answers.This relates to which guideline for asking questions?
Short items are best
treating missing data as one available response (IOW: make an assumption based on how certain questions were answered to fill in the missing data (like if they answered "yes" to many questions and left the ones that would make sense as "no" blank - assuming they mean "no")
a way to deal with the problem of missing data
decide how to score missing data after a careful analysis
a way to deal with the problem of missing data; clue: a study measuring religious beliefs found that people who answered "don't know" about a given belief were almost identical to the "disbelievers" in their answers about other beliefs.
The best solution to approach the problem of question ordering is sensitivity to the problem by attempting to estimate the resulting effect.
True
True or False: Data-collection instruments such as questionnaires are certain to contain error.
True
True or False: Like quantitative research, qualitative research often involves gathering data by directly observing people in addition to asking them questions
True
True or False: The opposite is generally true in terms of where to place different types of questions for interview surveys.
True
True or False: Sometimes a set of contingency questions is long enough to extend over several pages .
True
True or False: boxes adequately spaced apart are the best format for respondents
True
True or False: question order can affect the answers
True
if all but a few respondents respond to every item, exclude from the analysis the data from those few respondents whose index/scales are incomplete (IOW: if pretty much everyone completed their sheets and a few have missing data - just throw out those few)
a way to deal with the problem of missing data; clue: almost everyone completed all of their responses
"The state should abandon its community-based services and spend the money on improving institutional care." is an example of what?
a double-barreled question
avoid biased items and terms
a guideline to asking questions that is related to the idea that some questions seems to encourage particular responses more than other questions
interview guide
a list of topics to be asked about but not the exact sequence and working
questions that encourage respondents to answer in a particular way are called
biased
questionnaire items should be _ and _
clear and unambiguous
*In regards to questionnaires, whether to be completed by respondents or administered by interviewers, you should include _ _ and _ _ when appropriate
clear instructions and introductory comments (you should begin by telling them exactly what you want)
*The format of qualitative interviews varies how?
completely unstructured, informal conversational interviews - semistructured interviews (interviews guide outlining topics) highly structured, standardized interviews
The following are examples of what? marital satisfaction, level of social functioning, level of client satisfaction with services, quality of life
complex variables that social work researchers may find difficult to obtain adequately with a single questionnaire item
*variables may be too _ or _ to be measured with only one item
complex, multi-faceted
If question order is especially important then you might do what?
construct more than one version of the questionnaire - at the very least, you should pretest the different forms to measure ordering effect
What are the 2 key elements in the clearest and most effective format for contingency questions?
contingency question is set off to the side and enclosed in a box (isolated from the other questions) and an arrow connects the contingency question to the answer on which it is contingent
A guideline to asking questions: Questions should be _ sensitive
culturally (items that are clear in one culture may not be clear in another)
differences in scores on indexes and scales at the ordinal level of measurement ....
depict imprecise differences in quantity
"Should taxes be raised so welfare funding can be increased?" is an example of ...
double-barreled question
You should resist the temptation to insert these words when attempting to vary positively and negatively worded items
guideline for asking questions: avoid words like "no" or "not"
two structural requirements in the construction of closed-ended questions
exhaustive (included all possible responses that might be expected) and mutually exclusive (which means you can't have one and the other - choosing one will rule out the others) (the respondent shouldn't feel compelled to select more than one)
you should search for _ instruments that have already been tested with success and which fit your intended research. If they don't fit, you may only need to make minor modifications.
existing
*What are some ways to deal with the problem of missing data?
greatly depends on the research situation but ... 1. if all but a few respondents respond to every item, exclude from the analysis the data from those few respondents whose index/scales are incomplete (IOW: if pretty much everyone completed their sheets and a few have missing data - just throw out those few) 2. treating missing data as one available response (IOW: make an assumption based on how certain questions were answered to fill in the missing data (like if they answered "yes" to many questions and left the ones that would make sense as "no" blank - assuming they mean "no") 3. decide how to score missing data after a careful analysis 4.assign a replacement value to missing data: if you're creating an index out of several items, can use proportions based on what is observed 5.the safest and best method would be to construct the scale or index using alternative methods and see now each affects findings
among the most prominent techniques that social work researchers use to collect data
instruments designed to gather data by communicating with people orally or in writing (questionnaires, interview schedules, and scales)
Uniform scoring allowed by Likert-type response categories assumes that each item has about the same _ as the rest
intensity
One of the most important objectives in designing qualitative instruments is ....
probing for depth of meaning from the respondent's perspective
*both questions and statements may be used _
profitably
closed-ended questions are popular because...
provide greater uniformity of responses and are more easily processed
chief shortcoming of closed-ended questions ...
lies in the researcher's structuring of responses - in other words, you may miss something: what's the most import. problem facing your community? - you're checklist may omit some important problems that person thinks are important
in guidelines for asking questions, what are 2 characteristics of questions you should keep in mind
make items clear, avoid double-barreled questions
What two things are true about respondents in terms of following guidelines to asking questions?
must be willing to answer and must be competent
stylistic differences: quantitative approaches _ use interviewer flexibility regarding working, sequencing, and conversational style - qualitative approaches _ use them
never, very often
Complementary Functions: quantitative approach vs. qualitative approach
objectivity and consistency vs. subjectivity and flexibility
stylistic differences: quantitative approaches _ use questionnaires or scales - qualitative approaches _ use them
often, rarely
Indexes and scales typically provide _ _ of variables. In other words, they rank-order people or units of analysis in terms of specific variables such as attitude toward mother , social functioning and the like.
ordinal measures
a critical aspect of questionnaire design is
precoding - the info is typically transformed into some type of computer format so appropriate to include data processing instructions on the questionnaire itself - they indicate where specific pieces of info will be store in the machine readable data files
in guidelines for asking questions - identify the options of how to approach questions
questions and statements, open and closed-ended questions
open-ended questions
questions for which respondents are asked to provide their own answer, rather than selecting from among a list of possible responses provided by the researcher
contingency questions
questions that depend on the answers to previous questions, Subsequent questions are contingent on responses to the first question used to screen or limit responses to those who are qualified to respond
This technique should only be used when no other method will produce the desired result- if used, the list of answer categories should be relatively short
rank-ordering responses - bc may have to read and reread the list several times- involves time and can be difficult
stylistic differences: probes _ and _ for quantitative approaches _ and _ for qualitative approaches
rare and brief, frequent and in-depth
stylistic differences: quantitative approaches _ use open ended questions - qualitative approaches _ use them
rarely, usually
refer to pg220 "learning from bad examples" these could very likely be on the exam
refer to pg220 "learning from bad examples" these could very likely be on the exam
stylistic differences: formality of interview (description ) quantitative approaches (description )qualitative approaches
relaxed, friendly demeanor, but professional tone and not overly casual more likely to resemble a spontaneous, informal, friendly, conversation
Indexes and scales allow us to ....
represent complex variables with scores that provide greater potential for variance than would a single item. Analyzing one score derived form multiple items is also more efficient than analyzing each item separately.
closed-ended questions
respondent is asked to select an answer from among a list provided by the researcher
What is wrong with the following question: Report the age at which you first talked back to your parents.
respondent probably won't remember (aside from the problem of defining "talking back to parents")
asking a teenager how many miles they have driven since they got their license is an example of what guideline not being followed?
respondents must be competent to answer
The composite or cumulative measures of complex variables are called
scales or indexes
requests for duller demographic data might be placed at the end of a _
self administered questionnaire
both open and closed-ended questions may be used in what kinds of measurements?
self-administered questionnaires and interview schedules
stylistic differences: quantitative approaches _ use interviews - qualitative approaches _ use them
sometimes, usually
what are some guidelines for constructing questionnaires?
spread out and uncluttered,
stylistic differences between quantitative and qualitative approaches (p238chart)
stylistic differences between quantitative and qualitative approaches (p238chart)
nominal measures
such as gender or ethnicity, assess qualitative differences only, and don't convey more or less of something
among the most important objectives in designing quantitative instruments is....
the avoidance of measurement error (thus we seek to to construct instruments that are reliable and valid)
what is just as important as the nature and wording of the questions on a questionnaire?
the format
interview schedule
the list of questions asked by the interviewer
What kinds of questions would be best to begin a self administered questionnaire?
the most interesting set of questions **but should not be threatening
the rank order of ordinal measures is determined by what?
the overall score that combines all of the scale's items
the values of the likert scale format
the unambiguous ordinality of response categories ( easy to judge the relative strength of agreement intended by various respondents) lends itself to a straightforward method of scaled or index construction - each item can be scored in a uniform manner
The overall scores of the likert method are not the final product of the index or scale construction; rather, ....
they are used in an *item analysis* to select the best items
Although ordinal measures are very useful in quantifying variables....
they do not specify the precise degree of quantitative differences.
True or False: Introduce each section of questionnaire with a short statement about its content and purpose
true statement
ratio measures
unlike ordinal measures, have a precise mathematical meaning
stylistic differences: closed-ended questions: quantitative approaches use them _ qualitative approaches _ use them
usually, sometimes