Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
33 Cards in this Set
- Front
- Back
measurement
|
the process of assigning numerical values to phenomena in accordance with specific rules for representing quantities or qualities of attributes
|
|
the measurement process
|
id a concept
define the concept operational definition develop a measurement scale eval reliability and validity of the scale utilize the scale |
|
nominal scales
|
classifies subjects into groups/categories
typical app: often used to classify subjects into groups by sex, geog area, income range. -the numerical values have no true meaning w/the concept but simply rep labels to help classify statistics: frequencies/%ages, mode dichotomous - can't calc a meaningful mean |
|
ordinal scales
|
determines order but no magnitude of difference/distance
-shows position in relation to other options typical app: rankings/rank order statistics: all avail for nominal plus median (with specific formula) |
|
interval scales
|
rule: measures order and rel distance; intervals btwn adjacent ranks are assumed to be =
typical apps: likert scale, stapel scale, semantic differential scale statistics: all available for ordinal plus mean and std deviation |
|
ratio scales
|
rule: measures absolute order and absolute distance. there is a meaningful or absolute zero making comparison to the measurement to absolute possible
typical apps: sales figures, price paid, absolute income lvls (not ranges), absolute age statistics: same as interval |
|
uni-dimensional
|
designed to measure one attribute of a concept (loyalty)
|
|
multi-dimensional
|
designed to measure several dimensions of a concept (brand image)
|
|
itemized rating scale
|
can be used to measure:
satisfaction awareness frequency of use interest attitudes agreement with statements importance of attributes/features |
|
likert scale
|
special type of itemized rating scale
one of the most commonly used scales in mktg surveys always 'strongly disagree' and strongly agree anchors typically 5-7 response categories (need odd # for neutral/no opinion) |
|
rank order scale
|
can be problems if list isn't comprehensive
be careful of # of items asked to rank ordinal scale |
|
constant sum scale
|
works best with higher educated respondents
keep # of options to a max of 10 ratio data |
|
semantic differential scale
|
very common in mktg surveys, esp when measuring brand image
respondents often confused how to respond requires researcher to id bi-polar adjectives (often diff to be exactly opp) try to avoid 'halo effect' by rotating pos and neg sides interpreting the middle pt is diff (either neutral or unaware) |
|
stapel scale
|
overcome probs of sem. diff scale
overcome need to id bi-polar adjs; use only 1 consumer confusion about how to respond. provide very explicit, clear instructions w/ex |
|
purchase intention scales
|
very impt measurement when testing new product/service concepts
time frame is impt for any measurement of purchase intention often used for go/no go decisions for further R&D |
|
eval the reliability and validity of measurements
|
measurement = accuracy + error
accuracy would mean you have an accurate measurement of the concept of interest |
|
types of error
|
systematic error: constant error in the measurement process
random error: transient error in the measurement process (not consistent for all respondents) |
|
examples of systematic error
|
items included in the questionnaire were an incomplete rep of the concept of interest
measurement instrument was unclear/ambiguous problems w/the quality of the survey |
|
ex of random error
|
variations in admin the survey to each respondent
short term personal factors or situational factors for the respondent |
|
reliability
|
measurement scale that probides consistent results over time
ways to assess reliability -stability: test-retest; poss problems: opinions actually changed or first measure sensitized respondent internal consistency -equivalent form reliability most common -splitting halves |
|
validity
|
measures what it is intended to measure (no/little total error)
-face validity (whether measurements make sense logically) -content validity are the measurements used a complete measurement of the concept? -criterion rel validity: predictive validity (how good was it at doing what it was intended to do; only AFTER running survey |
|
issues to consider in selecting and designing measurement scale
|
type of scale
reliability and validity (pre test) number of scale categories balanced vs. unbalanced (heavily focused on one side) forced choice (odd/even # of scale categories) |
|
good questionnaires
|
ask the questions nec to achieve the research objs and test the hypoth
ask ?s that are high in validity and reliability -well phrased -approp seq -unambiguous -unbiased consider respondents ability to answer and understand ?s have been pre-tested |
|
questionnaire design process
|
id ?s
survey collection method ? response format ? wording flow and layout pre-test and revise final copy implement |
|
id ?s to ask
|
depends on objs and hypoth
multi item concept vs. single item filler questions |
|
choose the survey method
|
effect of method on ? format - can provide method
|
|
choose the form of ? responses
|
form of ?:
open ended close ended (nominal); dichot/mc closed ended scaled (interval) issues to consider: cost and time - ease of tabulating and respondent completing ease of respondent understanding |
|
choose the ? responses
|
open vs. closed ended mc
questions should be exhaustive and mutually exclusive avoid order bias: ? order bias, response order bias |
|
phrasing/wording of ?s
|
wording must be clear and unambiguous
avoid biasing the respondent -don't use a specific ex to measure a broader sit -avoid double barreled questions -avoid leading ?s -a series of likert scales is not considered leading consider the respondent's ability and willingness to answer the ? -no mktg jargon -use respondent's vocab -avoid making assumptions that aren't obvious -don't ask specifics when respondent likely to only remember generalities -consider respondent's willingness to answer; avoid loaded ?s |
|
determine flow and layout
|
1. proper title and intros
2. screener ? s to id qualified respondents 3. use the funnel technique -ask general ?s first -ask ?s that require work in the middle 4. position sensitive, threatening, and demographic ?s at the end 5. allow plenty of space for open-ended responses 6. instructions should be clear and easy to id 7. be aware of anchoring and order bias 8. each ? should begin and end on the same page 9. layout should be prof, not crowded, easy to read 10. for web, have a status bar 11. use a proper closing |
|
pretest
|
specific ?s for alternative forms, meaning and understanding (task difficulty)
pretesting the questionnaire: respondent interest and attention, flow of ?naire, skip patterns, length/time to complete |
|
rate of returned surveys
|
# completed/ # in sample =?
often low leads to biased results attempt to increase over 20% |
|
techniques for increasing
|
$$ incentive
prelim notification and follow up reminder effective cover letter w/personalization survey quality -length, reproduction, appearance, interesting questions, survey sponsorship or appeals |