click below
click below
Normal Size Small Size show me how
RMDA ch 7
| Term | Definition |
|---|---|
| what do surveys measure? | knowledge; attitudes and beliefs; behavior |
| knowledge | information (e.g. consumer choices); awareness (e.g. public health resources) |
| attitudes and beliefs | preferences or evaluations (e.g. attitudes towards specific groups); beliefs (abt political or social events); feelings or moods |
| behavior | behavioral intentions (e.g. intent to vote); self reports of ongoing or prior behavior (drug use etc) |
| important to define research questions | descriptive, test hypotheses, examine generalizability, predict event or outcome, pragmatic or applied |
| descriptive | epidemiology (e.g. drug use surveys); political/social description (e.g. opinion polls) |
| test hypotheses | assess blocking variables (e.g. recruiting for gender differences); correlational studies (what patterns go together) |
| pragmatic/applied | marketing, consumer surveys |
| constructing survey questions | simplicity; avoid double-barreled questions, avoid leading and loaded questions; avoid negative wording; avoid yea/nay saying |
| simplicity | avoid jargon; easily understood words |
| avoid double-barreled questions | don't ask two things at once (not simple) |
| avoid leading and loaded questions | questions should have neutral tone |
| avoid negative wording | avoid words like "not" |
| avoid yea/nay saying | don't have all answers be yes/no |
| question formats | close-ended; open-ended |
| close-ended | specific rating scale or highly structured prompts, most reliable for concrete behaviors; behavioral indicators |
| close-ended pro | clear operationalization; easy to quantify and use statistically |
| close-ended con | potential insensitivity (range of circumstances); superficial; not sensitive to participants; no option for "no opinion" |
| Likert rating scale | 5 or 7 point rating scale from strongly disagree to strongly agree |
| graphic rating scale | two endpoints and participant enters answer along line 100mm long; experimenter measures with ruler |
| somatic differential scale | examines respondents' meaning of concepts, series of bipolar adjectives |
| nonverbal (pictoral) scale | for children; range of faces for pain etc |
| open-ended | general textual/qualitative response; more sensitive to respondent; more difficult to interpret; can be mixed scales |
| if rating scale includes zero... | it's not likert! |
| forms of survey administration | self-report questionnaire; face-to-face interview; telephone interview; focus group |
| survey participants called | respondents |
| sampling error | deviate same amount from how the population would respond; not a mistake; some margin of error (aka confidence interval) |
| confidence level | how confident you are abt this value (e.g. 90% chance of rain, you're highly confident it'll rain) |
| confidence interval | what you calculate and report; report at 95% confidence; (e.g. 95% confident if we sampled pop, mean between 45-60) |
| confidence interval formula | Za/2 x SD/ sq root n; 1.96 x SD/ sq root n |
| issues in surveys: cost/population access | disenfranchised; lack of cell phones; stigmatized population; rural area |
| issues in surveys: participant sophistication | difficulties with affective reasoning; unable to accurately safely report; rationality bias |
| affective reasoning | predicting how you will feel at some point in future |
| rationality bias | assumption that ppl have rational reasons for doing things |
| issues in surveys: social desirability responding | high face-validity yield less valid responses; high social desirability wording; populations differ on social desirability |
| confound in studying differences between populations | populations differ on social desirability; e.g. male vs female sexual activity |
| social desirability minimized by | anonymous surveys, assurances of confidentiality, computer administration |
| issues in surveys: question order | questions trigger participants' memory or attention and can bias following questions (can be limited by counterbalancing) |
| issues in surveys: bias or fraud | political and cultural debates; pressure for confirmatory results encourages bias or fraud; bias in interpreting data |
| forms of sampling: probability (random sampling) | try to represent population, won't have equal number of participant categories; random selection; ideal |
| forms of sampling: non-probability sampling | use available samples for convenience; not random; externally valid to small/unusual populations |
| probability/random sampling | all pop equal chance of being sampled; choose participants in systematic random fashion |
| probability/random sampling pros | eliminates biases of convenience sampling |
| probability/random sampling cons | may under-sample unusual populations |
| non-probability sampling | useful for unusual pops; no sampling frame available (e.g. census data); likely to misrepresent pop, hard to detect |
| respondent driven sampling (RDS) | allows for "targeted" population estimates, aka snowball sampling (sensitive to incentives but access hidden pops) |