Lecture+7

•Unlike qualitative, involves metrics reduced to numbers •Not just descriptive (e.g., counting) but inferential (e.g., predictive, indication of correlation) •Controlled specific measurement of phenomenon •Not just X related to Y (correlation) but X causes Y (causation) - provided overarching theoretical model is correct of course •Examples of use in design? •Controlled environment •Precise responses, •Causality can be inferred
 * __ Class 7: Contextual Inquiry and Quantitative Methods __**
 * __ Quantitative Methods __**
 * __ Experiments __**
 * // Benefits and Limitations //**

•Must constrain outside influence •Can be hard to measure and determine causation in complex multivariate phenomena •Creates fake tasks in fake contexts - validity issues •Observation without being there - quantitative artifacts - e.g., Web access logs, click regions, turnstile counts, etc. •Records consequences of actual action •Examples of use? •Results based in actual action vs. perception or intent •Potential for complete sample •Once infrastructure set, data collection can be automatic
 * __ Data Mining __**
 * // Benefits and Limitations //**

•Privacy concerns - data could be used for less than noble purposes •Evidence of actions might not reflect motivations for why actions are done •Common method of obtaining information from broad cross-section of people •Quality of information directly depends on who is surveyed and the quality of questions asked •Online tools help - e.g., http://www.surveymonkey.com •Who is included in sample? •How are they reached? •Responders vs. non-responders - are they qualitatively different groups? •Sampling errors and confidence intervals (or, what +/- 3.5%, 19 times out of 20 means…) •Examples of sampling error •Understandable •Unambiguous •Data is relevant to research goals •Data can be easily analyzed •Limited in scope - take respondent’s attention span and willingness to help into account! •Specific better than general •Open/closed-ended questions - benefits and challenges •Opening and closing preamble and instructions important - especially if you’re not there to supervise •Test it before you use it! •Nominal (categorical differences; student, staff, faculty, etc.) •Ordinal (rank order implied; e.g., year of study) •Interval (ranks w/ equivalent intervals - e.g., grade point averages) •Ratio (intervals amenable to mathematical manipulations; e.g, age in years •Pick the right one for the question at hand! •1-5, 1-7, 1-9 scales •Midpoint - what does it mean? If no opinion, give that option •Can force decisions with even number of options (1-6) •Take care in too many consecutive items with same polarity of options - leads to patterned responses •Set of related questions measuring attitudes, beliefs, orientations etc. •Ex: multiple intelligences, personality tests (others?) •Scales must actually measure what they claim, not be redundant •Scale reliability - consistency in response patterns •Verify authenticity (esp. in web searches) - many scales are meaningless, published scales generally stronger •Simple description (55% of respondents are male, 75% under 25, etc.) •Cross-tabulation (males answered this, females answered that) •Correlation (direction and strength of link between two variables - slope and r-values) •Causation (inferential tests, confidence of variation) A **cross tabulation** (often abbreviated as **cross tab**) displays the joint distribution of two or more [|variables]. They are usually presented as a [|contingency table] in a [|matrix] format. Whereas a [|frequency distribution] provides the distribution of one variable, a contingency table describes the distribution of two or more variables simultaneously. Each cell shows the number of respondents who gave a specific combination of responses, that is, each cell contains a single cross tabulation. The following is a fictitious example of a 3 × 2 contingency table. The variable “Wikipedia usage” has three categories: heavy user, light user, and non user. These categories are all inclusive so the columns sum to 100%. The other variable "underpants" has two categories: boxers, and briefs. These categories are not all inclusive so the rows need not sum to 100%. Each cell gives the percentage of subjects who share that combination of traits.
 * __ Surveys/Questionnaires __**
 * __ Sampling __**
 * __ Question Guidelines __**
 * __ Question Guidelines (2) __**
 * __ Variable Data Types __**
 * __ Likert Scale Questions __**
 * __ Scales __**
 * __ Data analysis __**
 * __ Descriptive __**
 * __ Cross Tabulation __**
 * ||  || boxers || briefs ||
 * heavy Wiki user || 70% || 5% ||
 * light Wiki user || 25% || 35% ||
 * non Wiki user || 5% || 60% ||

from Wikipedia An indication that there is a relationship between two variables meaning that one effects the other Correlation’s can be positive or negative A positive correlation exists when an increase in the value of A increases the value of B A negative correlation exists when an increase in the value of A decreases the value of B
 * __ Correlation __**

correlation is not causation it shows thare is some relashonship but it is not nessasarly casue and effect

in the exsample "does TV watching cause higher perceived dangers in the world, or do paranoid people seek shelter from the world in their houses and end up watching more TV?" you can show a Correlation in this exsample but you cant say what is casusing what •Assuming causation can be dangerous in previous examples •Hard to prove causation from correlation - must tie to logical explanation of correlation (which can itself be debatable) •Ex: “mean-world” syndrome - does TV watching cause higher perceived dangers in the world, or do paranoid people seek shelter from the world in their houses and end up watching more TV?
 * __Causation__**?

•Social network questionnaire - who trusted whom in six domains •Correlated with three scales - interdependence, independence and proactivity •Correlational study - what relations existed between scales and position in network? •Causation - not easy to prove •Relations verified by respondent reflection and personal experience •Organizational redesign implications
 * __ Org. research example __**