Term | Notes |
Aided recall | a memory aid used to determine whether the respondent has seen something before - perhaps an advertisement |
Coding | In data analysis for research, the alternative responses are represented by codes, which then enable the statistical procedures to operate on simple data elements |
Correlation | the extent to which two variables are mutually dependent; see regression |
Counts | the sum of all responses to a single option of a question |
Cross-tabulation | The main form of reporting research in which the data are analysed in rows and columns and the counts and/or row and/or column percentages are reported. |
Dichotomous choice | a question where the respondent has to make a choice between one of only two alternatives |
Group discussion | a "meeting" of a number of respondents with a trained moderator to discuss a range of issues. Useful for eliciting ideas and concepts that might not otherwise emerge from an interview |
Multiple-choice | a question where the respondent may tick as many as apply |
Non-response | where the respondent has failed to complete any or all questions |
Open-ended questions | where the respondent is expected to write in or the respondents comments are captured verbatim. They may be reported verbatim, or post-coded |
Post-coding | where the data entry process creates codes to match the replies given, especially from open-ended questions where they may not be reported verbatim, but some sort of classification is required. |
Pre-coding | where the responses have been coded prior to the fieldwork so that the data can be captured quickly and efficiently |
Panel | where a specific sample has agreed to do research regularly |
Pilot study | where the survey is undertaken with a small sample with a view to identifying any weaknesses in the questionnaire, such as lack of clarity in questioning or insufficient pre-coded responses |
Population | the total number of people or things from which the sample is to be taken; also, Universe |
Quantitative research | research which is intended to measure the level or size something |
Qualitative research | research to determine respondent's attitudes or motivation |
Quartile | Four one-fourth parts of the sample in a cumulative frequency distribution |
Random sample | a sample drawn from the population such that each member of the populations has an equal chance of being selected |
Regression | the equation that describes how two correlated variables interact |
Respondents | that part of the sample that completed the survey |
Routing | directions given in a survey to go to a different part of the survey dependent upon the response |
Sample | a selected and representative proportion of the population that is to be researched |
Stratified sample | where the population is divided into segments, each of which is randomly sampled. This is useful to make sure that a sufficient number of responses are obtained from each segment for accuracy, although the results may be weighted back to the proper distribution of the segments in the sample |
Trend | the tendency for a variable to change over time |
Universe | the total number of people or things from which the sample is to be taken; also, Population |
Weighting | where the results are analysed while changing the relative importance of segments of the sample. This is used to correct an imbalance in the distribution of the responses compared with the population or to account for some other factor, such as sales potential in strategic research |
| |