The design of experiments (DOE) is the study of how to structure an information-gathering exercise where variation is present. In statistics, controlled experiments are usually implied, meaning that study units are randomly assigned to conditions where levels of a variable have been independently ...

learn more… | top users | synonyms (1)

0
votes
1answer
15 views

Unbiased estimate for treatment effect in block randomized experiment where probability of treatment varies by block

I want to analyze the results of a block randomized experiment. Within each block, units are randomized to treatment and control as if that block were a completely randomized experiment. Specifically, ...
0
votes
0answers
9 views

using Process plugin for multiples categorial X variables (experiment)

I'm trying to use Process plugin for my full factorial plan (for my thesis) and have some questions about it: - I have a 2x2x2 factorial design and want to know if it s possible to use Process marco (...
0
votes
0answers
16 views

Are multiple testing corrections necessary when validating against an already processed and published database?

Short version: Do I have to apply multiple testing corrections when analyzing a single variable (from a multidimensional dataset) I have previously found to be relevant and validated in the lab using ...
1
vote
0answers
16 views

Using Monte Carlo simulations with subsequent element removal

I'm attempting to build an evaluation set for a logistic regression classifier and I've run into a statistical problem. The study involves a very large population (G) that has two properties of ...
1
vote
1answer
32 views

Mixed factor both “within” and “between” subjects

I am wondering about experimental designs in which the levels of a factor are partly within and partly between subjects. Assume for concreteness that the experiment is as follows: The response ...
0
votes
2answers
58 views

Is logistic regression a valid way of analyzing A/B testing results?

I'm very new to the idea of A/B testing and I want to see if my train of thought here makes sense. Suppose that I run an experiment with two designs. I get two sets of resulting data, one for each ...
0
votes
0answers
10 views

Formula for variance of estimated variance components for experimental design

Consider a simple linear mixed model with one random effect. $Y_{ij}=\mu+\alpha_i+e_{ij}$ where $\alpha_i\sim N(0,\sigma^2_\alpha)$ for $i=1,...,I$, and $e_{ij}\sim N(0,\sigma^2_e)$ and $cov(\alpha,e)=...
0
votes
0answers
10 views

help on mixed models in spss

I am trying to analyze a complex data set and I got confused. Here is a brief explanation: Everyone has different beliefs, i.e., fate, luck, etc. I identified 5 belief types which I measure with some ...
-5
votes
0answers
25 views

Neural Chess, Neurons And Hidden layers! [closed]

Hi I am wanting to make a Gigantic Chess Neural Network, I was wanting to have 800 base Neurons that feed into the Hidden layers I also want to have 3 Hidden layers of 800 Neurons each at it's ...
0
votes
0answers
7 views

Repeat measures with missing data

Need guidance: I have repeat measures (5 successive measurements) of symptoms (categorical) & labs (continuous). N=809 patients. Outcome var is 1. composite outcome (sum of complications) ...
1
vote
0answers
14 views

Full-Factorial Design (3^2): Not sure about right method for analysis

I ran an experiment (online-survey) showing venture capitalists 9 fictional startups that only differed in factor 1 and factor 2. Both factors had three levels: Factor 1 = 50,250,450; Factor 2 = 0.5,1,...
2
votes
1answer
27 views

any difference between randomly selecting before or after an event?

Is there a mathematical difference between randomly selecting a population of users before an event vs after an event, for the purposes of experimentation? For example, let's say we want to split all ...
0
votes
0answers
11 views

R package for attribute availability discrete choice designs?

Is there an R package for discrete choice designs that can accommodate attribute availability? For example, systematically show only 6 out of 8 attributes in any choice set.
0
votes
0answers
4 views

Fractional Factorial Design with Cost objective

I am interested in generating D-optimal fractional factorial designs in R, but also selecting the design that optimizes a (non-linear) cost function. As a simple example, suppose you have three ...
0
votes
0answers
5 views

Balance order of non-independent treatments to the participants

In my study, I will apply 3 treatments to each participant. The 3 treatments are treatment A (visual stimulus), treatment B (gustatory stimulus), and treatment A and B combined (visual and gustatory ...
1
vote
0answers
54 views

Experimental Design: Split Plot with 2 whole plot factors and 1 subplot factor. Understanding Error Terms

I usually have a little bit better experimental design, but this experiment has already been planted and can't be changed. Instead of being a true split plot design, in which case I would use ssp....
0
votes
0answers
6 views

Experimentally testing classification models

I have implemented a machine learning algorithm (say, algorithm A) to perform binary classifications of fraudulent and non fraudulent clients. I am already using other algorithms (say, B and C) to ...
1
vote
1answer
22 views

Can you break up a pairwise comparison into components, so that individual participants don't have to assess every possible comparison?

I'm trying to do a pairwise comparison for 40 items. However, this results in $\frac{40 \times 39}{2} = 780$ combinations. Obviously no-one is going to sit through a survey that long. Is there a way ...
3
votes
1answer
21 views

What is the difference between a Within Subject and a Repeated Measure design

I am having some trouble understanding the difference (if any) between a Within Subject and a Repeated Measure design. A few resources online seem to suggest that they are often interchangeable or ...
0
votes
0answers
3 views

Design experiment to measure dependence of probability on a parameter

I have a physical system that can be modelled as following: there is a function $f:[0,1] \to [0,1]$ so that, given a parameter $\theta$, the physical system computes $p=f(\theta)$, samples $y$ from a ...
0
votes
0answers
24 views

How to compare experimental results of multiple models using significance tests?

I am using 5-fold cross validation to evaluate the errors of 10 models. Therefore, there are 50 results (each model has 5 error values generated from 5 random evaluation partitions). My question is: ...
0
votes
0answers
8 views

Resources for Determining Value-Added by Parameter Sweeps

I am working on a project that aims to use Design of Experiments to improve our testing process. We are performing tests using a parameter sweeps of numerous input variables (40+). Each input variable ...
0
votes
1answer
18 views

2 categorical IV and 1 ordinal DV — what test to use?

I'm a complete newbie when it comes to statistics, and I'm struggling to decide on the design for a statistics test, so I was wondering if someone might be willing and able to help me out please? ...
3
votes
2answers
98 views

Working with percentages of positive variables

Say we run an experiment and observe the following impact on a variable of interest (one row per experimental unit): ...
0
votes
0answers
27 views

Analysis of reaction times with missing data by trial/item

I have a developmental psychology study with one between factor (age) and two within factors (scale-3 and difficulty-2). I recorded reaction times but have eneded up with a lot of missing data for the ...
0
votes
0answers
5 views

How to estimate the concomitant variation

Is there a formula/sofwate solution to estimate the concomitant variation for a 3 factorial experimental design with one dependant variable?
0
votes
0answers
9 views

Account for Repetition in Experimental Design when testing variants of an image

In my experiment, an image is presented in five different ways to the test person. The person then has to do some task like "point to the red flower in the image" and I measure the time taken to check ...
4
votes
1answer
43 views

What is cross-lagged panel design?

The slides for my statistics lectures explain various experimental designs such as single-strata, cross-sectional and longitudinal. Unfortunately, the diagram for Cross-lagged Panel Design is very ...
1
vote
0answers
23 views

Same mean, different variance - test?

I have conducted an experiment where I am interested in how well different models for directional hearing works. For each model a sound is presented to a test subject, and the test subject expresses ...
1
vote
1answer
20 views

2X2 ANOVA: low power, statistical significance for only 1 IV, medium-low effect size

BACKGROUND: I run a 2X2 Between-Subjects ANOVA. I obtained: a low power (< 0.30) for both Independent Variables a not statistically significant result for one of the Independent Variables. ...
0
votes
0answers
12 views

Mean sum of squares and products according to source of variation in ANCOVA from R

How to get the mean sum of products using R to populate an ANCOVA table as given in the following link? Analysis of Covariance Table for RBD Ryy, Tyy and Eyy can be obtained as follows using ...
0
votes
0answers
15 views

Analysis of measures of difference

Let's say I want to measure the width of a road using a measuring tape. I've drawn a crap diagram here: Rather than measure the distance directly, I decide to be obtuse and measure the distances ...
2
votes
0answers
25 views

Applicability of survival analysis (using Cox proportional hazards) to question of interest

I have started down the path of using cox PH models to try to understand which variables are driving time-to-migration (event) in my study system. My system has two groups (A,B) of animals that ...
0
votes
0answers
20 views

Multivariate GLM or MANOVA or other alternative for 2x2 repeated measure experiment?

I am trying to get my head around the right choice of experimental design and analysis for a collaborative software I have created and would like to evaluate. The software allows groups to conduct a ...
0
votes
0answers
30 views

Interference in experiment with block design

I'm designing an ecological experiment to test how 2 treatments (one manipulated, and one non-manipulated control) effect fish density at at their respective locations. The experiment will be done in ...
0
votes
0answers
18 views

Two nested random factors, planing an experiment

I'm planning an experiment (described below), and I want to be sure that the experimental design is correct, and I'll have no trouble later on when analyzing the data. I want to know if the ...
0
votes
0answers
11 views

How to design experiment with 1 IV (2 levels) and 2 DV's

My groups are control and alcoholics. I want to measure them based on an experiment that has 2 conditions: go (press button after x when you see o followed by x) and no go(when o appears but any ...
4
votes
1answer
110 views

outlier detection: area under precision recall curve

I would like to compare outlier detection algorithms. I am not sure if area under roc or under precision recall curve is the measure to use. A quick test in matlab gives me strange results. I try to ...
0
votes
0answers
13 views

Convenient plot size detectable difference vs Least Signifcant Difference

I am curious how Hatheway's (1961) Convienient plot size calculation, once rearranged to solve for the detectable difference, relates to a standard LSD calculation: $$ d^2 = \frac{(t_1 + t_2)^2 \times ...
2
votes
0answers
25 views

Estimate causal effect with lots of missing values

This is a generic question. Suppose you run an experiment with units randomized between a treatment and a control group. Imagine the response rate is very low in each group, say, below 10% and with a ...
1
vote
0answers
21 views

Repeated measures anova in R

I have a dataset containing percentage scores for 15 participants. There are two observations for every participant, one with an intervention and one with a placebo in a double-blind design. Each ...
0
votes
0answers
30 views

How to deal with statistically insignificant factors?

I used $2^4$ factorial design to screen significance of the factors $A$, $B$, $C$ and $D$. It turned out that $A$ and $B$ were significant, whereas $C$ and $D$ were not. Since there is no need to ...
0
votes
0answers
7 views

Expanding an experimental design

I have completed test work following a randomised response surface (multi-factor) experimental design (Central Composite Full) but have found that my levels for two of the factors are not broad enough....
1
vote
1answer
33 views

How to measure extraneous and confounding variables?

The test-re/post/test methodology entails establishing a baseline, introducing a condition and measuring the effect size. The condition is known. But whether or not it yielded some changes, is what ...
0
votes
0answers
8 views

Designing paired samples of Group A vs. Group B+C

I have decided to use a strict bioinformatics algorithms that under its way calculate fishers exact test. My samples are to parentals A1 and A2. These were treated and subgrouped into 14 clones. Two ...
0
votes
0answers
20 views

What family of GLM when response is Bray-Curtis dissimilarity? Should I use “adonis” instead?

I have some questions about testing for effects of different experimental levels on community similarity. I'll explain my planned experimental design before I ask the questions: I have two methods ...
0
votes
0answers
7 views

General strategy of confounding design

Example: Suggest a confouding scheme for a $2^8$ experiment in 16 blocks of 16, assuming that all 2-factor and 3-factor interactions are to be estimated. Please find the confounding design. Question: ...
0
votes
0answers
6 views

Completly Randomized Trials versus Incomplete Cubic Lattice

I have a 2x2x2x2 design, which I wish to balance. Previously, the binary factors appeared throughout the experiment with a 50% probability, which each have to result in a uniform distribution. I am ...
0
votes
0answers
17 views

sample size for multi-arm randomised control trial measured over time

I would really appreciate any advice on calculating a sample size: This study involves looking at a specific disease, then performing a randomised control trial in two different geographical areas ...
2
votes
0answers
45 views

What is the relationship between LATE and TOT?

My understanding of LATE was that it was the effect of a treatment on individuals who were induced to be treated by the experiment. That is, the effect on compliers. My understanding of Treatment-on-...