problem definition choose research design sample and collect data analyse data write report
step 3 exploratory?
>interviews, focus groups, case studies of groups or individuals -chosing numbers depends on the problem, target audience, time and cost
1) nominal, delphi, leaderless discussion 2) focus group discussion led by moderator with 8-12 homogenous people
step 3 descriptive?
>survey, secondary data 1) format 2) wording 3) content 4) sequence and layout
1) question format- structure/ disguise 2) question wording- simple, unambiguous, length 3) question content- one of more? required? 4) sequence and layout- easy first, sensitive last, general to specific
types of surveys? 5
personal phone mail online combo
influences on survey types? 4
target cost time response rate
measurement and scaling impacts on?
step 4 causal and descriptive? 3
>statistical techiniques to investigate relationship between: 2+ catagorical variables 2+ scale variables 1+ scale independent and 1+ scale dependant
sequential variables that follow numerical order
variables with no numerical relationship
2+ categorical variables tool? 2C CT
2+ scale variables tool? 2S C
1+ scale independent and 1+ scale dependent tool? 1S1C R
cross tab? frequency count
-combined frequency count -calc %s in direction of the IV -look for large %s along diagonal <50% -observational only >Chi Square
statistical test of the relationship between variables -sig <.05 relationship exists
correlation? linear relationship
-looks for linear relationship -<.05 shows shows 2 variables related -pearsons indicated strength of relationship (ignore negatives) 1=perfect linearity, 0=random -direction (+/-) -infer cause and effect e.g. draw causal diagram
regression? linear of IV and DV
-looks for linear relationship -<.05 shows relationship between IV and DV -regression equation to predict DV's for varying IV's -RSQ=variance in the DV explained by the IV's -backward regression=determine optimal V's to explain varience in DV
cross tab relationship exists when?
significance is <.05
if a relationship exists the null hypothesis is?
conjoint analyse variables?
scale dependent vairable and 1 or more independent variables that are levels (0 or 1)
0 or 1 e.g high or low
total of all available options in terms their attribute levels e.g 0,1,1
attribute dollar value?
price customers would be prepared to pay - (price level difference/ Price B) x Attributes B
assigning numbers/symbols to object according to pre-specified, uniform rules that do not change
scaled independently -interval or ratio scaled e.g. continuous/itemized rating scale- likert, semantic, stapel
numbers serve as labels for classifying -mutually exclusive e.g male=1 female=2 -based on frequency counts
categorises the variables in meaningful way -based on centile statistics
measures distance between any two points on scale -random start and measurement -based on additional statistical techniques e.g likert
all properties of NOI -has absolute zero -all statistical techniques
assigning a figure of importance to each attribute to total 100
agreement or disagreement to a statement about the stimulus
oxymoronic words on a scale
intensity and direction relating to a stimuli aspect
issues with noncomparative scales?
labeling ignorance balance
are you measuring what you set out to measure
consistent measures across time and items -internal consistency- highly correlated -stability- same results over time
reliability x validity
generalisation over all situations
validity and reliability?
1) V = R - no random/ systematic error 2) unR # V 3) R #/= V due to systematic error 4) R is neccessary but not sufficient condition for V
survey design? I M Q S W SL
1) info needed- exploratory research to gain insight- who? where? how? -consider target market 2) method 3) questions -that overcome unwillingnes to respond -avoid double barrelled questions -link to research 4) structure-open vs close ended, structured 5) wording-simple, appropriate length, non leading 6) sequence and layout- gernal to specific, senstive questions later
overcoming unwillingness to respond?
disguise e.g. phrasing in third person, using innocuous questions
factors affecting survey method?
sampling population type question form & content response rate costs staff time
personal telephone mail fax online combos
double barrelled question?
two or more questions combined into one -always break into two questions
inability to answer?
-uncertainty- provide alternatives -avoid generalisations and estimates
open ended questions?
no reponses given- survey intros, to many possibilities, precision is important -ad: no influence, wide range responses -disad: time consuming, variability of depth and response, requires interpretation, recording precision difficult
closed ended questions?
list or rating scale alternative responses provided -ad: les potential error, comparable responses, easy answers for interviewer -disad: meaningless results, limited self expression, reponse creation (bias/limitation)
structure and disguise?
1) SUD: do you..oxymoronic? 2) SD: do you..alternatives? 3) UUD: do you? 4) UD: projective techniques -associaiton, completion, construction, expressive